Method for acquiring maculopathy edema characteristics in retina image and electronic equipment

Abstract

The embodiment of the invention provides a method for acquiring maculopathy edema features in a retina image and electronic equipment, and relates to the technical field of image recognition. The method for acquiring the maculopathy edema features in the retina images comprises the following steps: acquiring a current retina image to be recognized in a plurality of retina images; determining an edema brightness threshold value of the current retina image based on the brightness value of the pixel in the vitreous body region of the current retina image; based on the brightness value of each pixel in the current retina image and an edema brightness threshold value of the current retina image, an edema area in the current retina image is determined, and the brightness value of the pixel in the edema area is smaller than the edema brightness threshold value. According to the invention, feature extraction can be carried out on the macular oedema region in the image based on the brightness value of the pixel in the retina image.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to a method for acquiring macular lesion edema features in retinal images and an electronic device. Background Art

[0002] The retina is a thin layer of cells located at the back of the eye. It is composed of photoreceptor cells (photoreceptors) and pigment epithelial cells. After being stimulated by light, the retina converts light signals into nerve signals and transmits them to the brain so that people can see. The retina is a very sensitive, shallow and complex structure, composed of photoreceptor cells, bipolar cells, ganglion cells, etc. Not only will ophthalmic diseases be reflected in the state of the retina, but some metabolic-related physical and mental diseases will also cause changes in the retina, such as diabetic retinopathy.

[0003] In the retina, the macula is an important area of ​​the retina and plays a vital role in vision. Due to its characteristics of containing a large number of cone cells, the macula is the most sensitive part of the vision, and the health of the macula will directly affect vision. Macular degeneration is a common retinal disease, including age-related macular degeneration, macular edema, etc. These diseases can seriously damage central vision and affect the patient's daily life. For macular diseases, ophthalmologists need to use advanced examination equipment and techniques, such as optical coherence tomography (OCT) and fundus fluorescein angiography, to achieve early diagnosis and precise treatment of the disease.

[0004] The current diagnosis of macular diseases is basically performed by well-trained radiologists who perform eye examinations. From the images obtained, experienced ophthalmologists read the collected images to draw conclusions on whether the patient has macular disease. However, the serious shortage of ophthalmic clinicians and radiologists is far from meeting clinical needs, and the experience accumulation and film reading ability of young ophthalmologists also require long-term training and cultivation. Through manual reading, images with pathological tissues are selected from large data sets, and diagnostic conclusions are drawn from subtle tissue lesions in the images. This traditional method of diagnosing diseases cannot serve daily medical needs in a large-scale, precise, and personalized manner. It cannot accurately identify, automatically analyze, and quantitatively track and evaluate the lesion site, and the goal of future precision medicine cannot be achieved.

[0005] At the same time, due to the rapid development of OCT technology, its clinical promotion and application have become more extensive, which has led to a large and rapid accumulation of OCT image data. With the iteration of technology, the image quality, imaging range and resolution presented by OCT have also been greatly improved. The diagnostic analysis software currently available for OCT images in clinical practice can only roughly capture and evaluate the central thickness of the macula, and the automatic recognition of retinal layers cannot achieve effective accuracy, which also leads to poor accuracy in the recognition of central thickness of the macula; it is even more impossible to accurately identify lesions, accurately identify the area of ​​lesion tissue for lesions, and perform quantitative statistics on important indicators such as edema area and subretinal effusion area. It is also impossible to objectively monitor and record the specific changes in the size of lesions during patient follow-up, not to mention obtaining effective change data through patient follow-up and treatment methods to support or predict the results of prognosis. The current dilemma not only brings certain troubles to the in-depth research in the field of macular diseases in the future, but also cannot meet the objective quantitative analysis of patients' conditions, and cannot make a good transition from qualitative to quantitative refinement, so it cannot help precision medicine.

[0006] Therefore, if it is possible to intelligently identify and quantitatively analyze macular lesions, and to conveniently process the acquired OCT images through an effective computer-assisted image analysis system to provide clinicians with quantitative and objective measurements, it will be of great help to doctors in making clinical decisions, predicting disease prognosis, analyzing disease progression through big data, and formulating accurate and personalized treatment plans for patients.

[0007] However, the accuracy of the diagnostic analysis software currently available for OCT images in clinical practice is often limited by the turbidity of the refractive medium and the ability of technicians to capture images, and is unable to effectively provide objective quantitative results of lesions. In addition, the description and quantification of changes in the lesion area through central thickness is not completely consistent and cannot fully explain the nature of the changes in the lesions.

[0008] Based on the rapid development of artificial intelligence technology, it is now possible to automatically identify and mark the lesion area through artificial intelligence by trying to input a large number of manually labeled OCT images, which can then be used to collect quantitative data of the lesions. Most commonly used macular disease recognition systems rely on machine learning models, which are trained by annotating a sufficient number of labeled OCT images to obtain a machine learning model with a high theoretical accuracy.

[0009] But this is only in theory, and there are many problems in achieving it: for example, machine learning training requires a large number of annotated OCT images, but the annotation of OCT images requires specially trained ophthalmologists to perform the operation, and professional ophthalmologists often do not have a lot of time to focus on the annotation of OCT images. In this case, it is very difficult and time-consuming to obtain high-precision annotation data, and it requires too high labor costs, which is one of the barriers to the use of artificial intelligence technology for OCT images. In addition, the lesions of different diseases are different, so they need to be trained separately. If the same model is used for lesion identification of all diseases, the accuracy of lesion identification will be too poor, which undoubtedly further increases the difficulty of artificial intelligence for OCT image recognition. In addition, machine learning models have high requirements for training and testing data. If the model encounters special cases or the OCT shooting machine is replaced during use, the robustness and accuracy of the trained model will also be reduced.

[0010] The above-mentioned problems show how difficult it is to apply artificial intelligence technology to OCT image recognition. In addition, even if artificial intelligence technology is applied to OCT image recognition, it is still unknown whether it can achieve the desired recognition effect. Summary of the invention

[0011] The purpose of the present invention is to provide a method for acquiring edema features of macular lesions in retinal images, an electronic device and a storage medium. The method can determine the edema brightness threshold of the retinal image based on the brightness values ​​of pixels in the vitreous area of ​​the retinal image, and then perform a binary judgment of edema and noise on the pixels in the retinal image based on the edema brightness threshold, thereby removing non-edema noise in the retinal image and realizing feature extraction of macular edema in the retinal image; that is, based on the information at the pixel level of the retinal image, automatic feature extraction of macular edema can be performed, and the macular edema area of ​​the retinal image can be obtained by using only a computer through automatic encoding and compilation technology. Compared with macular edema recognition by machine learning, it does not require training through data or high-performance hardware, and is simpler and easier to use; it achieves a higher signal-to-noise ratio, has the characteristics of high accuracy, fast recognition speed and low hardware requirements, and is suitable for a variety of macular edema feature extraction scenarios.

[0012] To achieve the above-mentioned objectives, the present invention provides a method for acquiring edema features of macular lesions in retinal images, comprising: acquiring a current retinal image to be identified from multiple retinal images; determining an edema brightness threshold of the current retinal image based on the brightness values ​​of pixels in the vitreous region of the current retinal image; determining an edema region in the current retinal image based on the brightness values ​​of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, wherein the brightness values ​​of the pixels in the edema region are less than the edema brightness threshold.

[0013] The present invention also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for acquiring macular lesion edema features in retinal images as described above.

[0014] The present invention also provides a computer-readable storage medium, which is a non-volatile storage medium or a non-transient storage medium, on which a computer program is stored, and is characterized in that when the computer program is run by a processor, it executes the method for acquiring macular lesion edema characteristics in retinal images as described above.

[0015] In one embodiment, the method further comprises:

[0016] Performing a plurality of preprocessing operations on the current retinal image to obtain a plurality of reference retinal images after the preprocessing operations;

[0017] Determining an edema area in each of the reference retinal images based on a brightness value of each pixel in each of the reference retinal images and an edema brightness threshold of each of the reference retinal images;

[0018] Based on the edema area of ​​each of the reference retinal images and the edema area of ​​the current retinal image and preset weight parameters, the effective edema area in the current retinal image is determined, and the preset weight parameters include: the weights of each of the reference retinal images and the current retinal image.

[0019] In one embodiment, before determining the edema area in the current retinal image based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, the method further includes:

[0020] Identifying the RPE layer of the current retinal image based on the brightness value of each pixel in the current retinal image;

[0021] Determining an edema area in the current retinal image based on a brightness value of each pixel in the current retinal image and an edema brightness threshold of the current retinal image includes:

[0022] Based on the RPE layer of the current retinal image, the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, the edema area in the current retinal image is determined, and the pixels in the edema area are all located above the RPE layer.

[0023] In one embodiment, determining the edema brightness threshold of the current retinal image based on the brightness values ​​of pixels in the vitreous region of the current retinal image includes:

[0024] Obtaining an initial brightness threshold value of a brightness value of pixels greater than a preset percentage of all pixels in the vitreous region, wherein the preset percentage is greater than 50%;

[0025] Searching for target pixels whose brightness values ​​are within a set brightness value range in the vitreous region, and counting the number of target pixels of each target brightness value based on the target brightness value of each target pixel; wherein the set brightness value range is a brightness value range with the initial brightness threshold as the midpoint;

[0026] Finding a target brightness value with the smallest target pixel number difference from the initial brightness threshold among all the target brightness values ​​as the minimum change brightness threshold;

[0027] The edema brightness threshold is determined based on the initial brightness threshold and the minimum change brightness threshold.

[0028] In one embodiment, determining the edema brightness threshold based on the initial brightness threshold and the minimum change brightness threshold includes:

[0029] The product of the average value between the initial brightness threshold and the minimum change brightness threshold and a preset correction coefficient is obtained as the edema brightness threshold, and the preset correction coefficient is greater than 0.5.

[0030] In one embodiment, determining the edema area in the current retinal image based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image includes:

[0031] In the closed area formed by pixels whose brightness values ​​in the current retinal image are less than the edema brightness threshold, a closed area whose area meets a preset condition is selected as an edema area;

[0032] The preset condition is that the area of ​​the closed area is smaller than the product of the area of ​​the effective area in the current retinal image and a set threshold and is larger than the area of ​​a preset number of pixels, and the set threshold is greater than 0 and less than 0.5.

[0033] In one embodiment, after determining the edema area in the current retinal image based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, the method further includes:

[0034] For each of the edema regions, the area of ​​the edema region is determined based on the pixel spacing information of the current retinal image and the number of pixels in the edema region.

[0035] In one embodiment, after determining the area of ​​each edema region based on the pixel spacing information of the current retinal image and the number of pixels in the edema region, the method further includes:

[0036] Based on the area of ​​each edema region in the current retinal image and the image spacing information of the current retinal image, the volume of each edema region in the current retinal image is obtained.

[0037] In one embodiment, before determining the area of ​​each edema region based on the pixel spacing information of the current retinal image and the number of pixels in the edema region, the method further includes:

[0038] Performing etdrs partitioning on the current retinal image to obtain the edema area within the etdrs partitioning;

[0039] For each of the edema regions, determining the area of ​​the edema region based on the pixel spacing information of the current retinal image and the number of pixels in the edema region comprises:

[0040] For each of the edema regions within the etdrs partition, the area of ​​the edema region is determined based on the pixel spacing information of the current retinal image and the number of pixels within the edema region.

[0041] In one embodiment, identifying the RPE layer of the current retinal image based on the brightness value of each pixel in the current retinal image includes:

[0042] Searching for a designated pixel with the highest brightness in a valid area of ​​the current retinal image by column, and for a designated pixel in a current column in the valid area of ​​the current retinal image that has been searched, if a longitudinal position difference between the designated pixel in the current column and the designated pixel in a previous column adjacent to the current column is greater than a first difference threshold, determining a longitudinal position of the designated pixel in the current column based on the longitudinal position of the designated pixel in the previous column adjacent to the current column;

[0043] The RPE layer of the current retinal image is obtained by combining the designated pixels in each column within the effective area of ​​the current retinal image.

[0044] In one embodiment, searching for target pixels with the highest brightness in columns within the effective area of ​​the current retinal image includes:

[0045] Starting from any middle column in the effective area of ​​the current retinal image, search leftward and rightward respectively for the target pixel with the highest brightness in each column.

[0046] In one embodiment, after obtaining the RPE layer of the current retinal image by combining the designated pixels in each column within the effective area of ​​the current retinal image, the method further includes:

[0047] Acquire a plurality of designated retinal images associated with the current retinal image, and determine a pixel reference height of the RPE layer based on an average height of pixels included in the RPE layer of the plurality of designated retinal images;

[0048] If the difference between the average pixel height of the pixels included in the RPE layer of the current retinal image and the pixel reference height is greater than a second difference threshold, the RPE layer of the current retinal image is determined based on the longitudinal position of the pixels included in the RPE layer of the specified retinal image adjacent to the current retinal image.

[0049] In one embodiment, the vitreous area of ​​the current retinal image is determined as follows:

[0050] The effective area of ​​the current retinal image is sampled in sequence using a sampling frame of a set size, and if the average brightness value of the pixels in the sampling area currently selected by the sampling frame is within a preset brightness range, it is determined that the sampling area currently selected by the sampling frame is the vitreous area. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a schematic diagram of a method for acquiring macular lesion edema features in a retinal image according to a first embodiment of the present invention;

[0052] Figure 2 is a schematic diagram of an original retinal image according to a first embodiment of the present invention;

[0053] Figure 3 yes Figure 2 Schematic diagram of the hierarchical structure of the retinal image;

[0054] Figure 4 yes Figure 1 A specific flow chart of step 102 of the method for acquiring macular lesion edema features in a retinal image;

[0055] Figure 5 is a schematic diagram of an edema area identified in an original retinal image according to the method for acquiring macular lesion edema features in a retinal image according to the first embodiment of the present invention;

[0056] Figure 6 is a schematic diagram of the current retinal image being partitioned by etdrs according to the first embodiment of the present invention;

[0057] Figure 7is a schematic diagram of a method for acquiring macular lesion edema features in a retinal image according to a second embodiment of the present invention;

[0058] Figure 8 is a schematic diagram of the current retinal image before and after non-local mean processing and sharpening processing according to the second embodiment of the present invention;

[0059] Fig. 9 is a schematic diagram of the current retinal image before and after automatic contrast adjustment processing according to the second embodiment of the present invention;

[0060] Fig.10 is a schematic diagram of the current retinal image before and after sharpening processing according to the second embodiment of the present invention;

[0061] Fig.11 is a schematic diagram of the current retinal image before and after being processed by the Otsu method according to the second embodiment of the present invention;

[0062] Fig.12 is a schematic diagram of a method for acquiring macular lesion edema features in a retinal image according to a third embodiment of the present invention;

[0063] Fig.13 is a schematic diagram of a portion of a current retinal image located below the RPE layer before and after being adjusted to black according to a third embodiment of the present invention;

[0064] Fig.14 It is a schematic diagram of an edema area identified in a current retinal image in which the portion below the RPE layer is adjusted to black according to the method for acquiring macular lesion edema features in a retinal image in the third embodiment of the present invention. DETAILED DESCRIPTION

[0065] The following will be described in detail with reference to the accompanying drawings to provide a clearer understanding of the purpose, features and advantages of the present invention. It should be understood that the embodiments shown in the accompanying drawings are not intended to limit the scope of the present invention, but are only intended to illustrate the essential spirit of the technical solution of the present invention.

[0066] In the following description, certain specific details are set forth for the purpose of illustrating various disclosed embodiments to provide a thorough understanding of the various disclosed embodiments. However, those skilled in the relevant art will recognize that the embodiments may be practiced without one or more of these specific details. In other cases, well-known devices, structures, and techniques associated with the present application may not be shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

[0067] Unless the context requires otherwise, throughout the specification and claims, the word "comprise" and variations such as "include" and "have" should be construed in an open, inclusive sense, ie, should be interpreted as "including, but not limited to."

[0068] References throughout the specification to "one embodiment" or "an embodiment" indicate that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of "in one embodiment" or "in an embodiment" in various places throughout the specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any manner in one or more embodiments.

[0069] As used in this specification and the appended claims, the singular forms "a", "an", and "the" include plural referents unless the context clearly dictates otherwise. It should be noted that the term "or" is generally employed in its sense including "or / and" unless the context clearly dictates otherwise.

[0070] In the following description, in order to clearly show the structure and working mode of the present invention, many directional words will be used for description, but the words "front", "back", "left", "right", "outside", "inside", "outward", "inward", "up", "down", etc. should be understood as convenient terms and should not be understood as restrictive terms.

[0071] The first embodiment of the present invention relates to a method for acquiring macular edema features in a retinal image, which is applied to an electronic device. The electronic device may be a common computer device, such as a desktop host, a laptop computer, etc., or a mobile phone.

[0072] like Figure 1 As shown, it is a specific flow chart of the method for acquiring macular lesion edema features in retinal images of this embodiment.

[0073] Step 101, obtaining a current retinal image to be identified from a plurality of retinal images.

[0074] Specifically, the retinal image is an OCT image of the user's fundus collected by an OTC device (i.e., an optical coherence tomography scanner). The computer device can be directly connected to the OCT device to obtain the DICOM format file sent by the OCT device, or it can be connected to other intermediate data collection devices or storage devices to obtain the DICOM format file collected by the OCT device; the DICOM format file includes the retinal OCT image and various other information related to the retinal OCT image, such as: shooting date, serial number, patient ID, device serial number, shooting date, pixel spacing, image spacing, etc. The computer device can parse the acquired DICOM format file through DICOM file recognition software to obtain the retinal image and the information of each retinal image. The retinal image can be stored in a common format (such as JPG, PNG, etc.) to facilitate subsequent image processing; the information of each retinal image is converted into elements that can be accessed through the tag name. Such as Figure 2 The figure below shows a retinal OCT image extracted from a DICOM format file, which is the original retinal image. Figure 3 , is a schematic diagram of the layered structure of the eye in the retinal image, from top to bottom: ILM layer (Internal Limiting Membrane), RNFL layer (Retinal Nerve Fibre Layer), GCL layer (Ganglion Cell Layer), IPL layer (Inner Plexiform Layer), INL layer (Inner Nuclear Layer), OPL layer (Outer Plexiform Layer), ONL layer (Outer Nuclear Layer), ELM layer (External Limiting Membrane), PR layer (Photoreceptor Layers), RPE layer (Retinal Pigment Epithelium), BM layer (Bruch's Membrane), CC layer (Choriocapillaris), CS layer (Choroidal Stroma).

[0075] Therefore, the multiple retinal images acquired by the computer device are obtained by slicing a three-dimensional stereoscopic image of the user, and image recognition is performed on these multiple retinal images in sequence. After the recognition of each retinal image is completed, the next retinal image to be identified (i.e., the next image obtained by slicing), i.e., the current retinal image, is determined, and the brightness value of each pixel in the current retinal image can be read out first. Among the multiple retinal images obtained after slicing a three-dimensional stereoscopic image of the user, the image spacing between any two adjacent retinal images is also known, and the image spacing can be a fixed value or a non-fixed value, that is, the image spacing between any two adjacent retinal images can be the same or different.

[0076] Step 102: Determine an edema brightness threshold of the current retinal image based on the brightness values ​​of pixels in the vitreous region of the current retinal image.

[0077] Specifically, based on the similarity between the brightness of the macular edema area and the brightness of the vitreous area in the retina, the brightness threshold of the macular edema area can be determined with reference to the brightness of the vitreous area in the retina, which is recorded as the edema brightness threshold.

[0078] The vitreous region in the current retinal image is the region on the ILM layer. The specific method for determining the vitreous region of the current retinal image is as follows:

[0079] First, the effective area of ​​the current retinal image is obtained to remove the black or white border caused by the edge of the current retinal image exceeding the imaging area during OCT shooting.

[0080] The specific process is to search for the black or white border that exceeds the imaging area row by row and column by column starting from the edge of the current retinal image to obtain the effective area of ​​the current retinal image; taking the upper edge of the current retinal image as an example, starting from the first row of the upper edge, calculate the average brightness of all pixels contained in the row. If the average brightness value of the row is greater than 50 or less than 5, then determine that the row is a black or white border that exceeds the imaging area, and continue to determine whether the average brightness of all pixels in the second row is greater than 50 or less than 5, until reaching the row where the average brightness of the pixels is not greater than 50 or less than 5. Similarly, the rows and / or columns where the four edge portions of the current retinal image exceed the imaging area (i.e., the effective area) can be determined. These rows and columns form the border of the effective area, and the image within the border is the effective area of ​​the current retinal image.

[0081] Then, the vitreous area is determined in the effective area of ​​the current retinal image. The determination method is: use a sampling frame of a set size to sample the effective area of ​​the current retinal image in sequence, and if the average brightness value of the pixels in the sampling area currently selected by the sampling frame is within a preset brightness range, then the sampling area currently selected by the sampling frame is determined to be the vitreous area; the sampling frame can be set as needed, generally a square or rectangular frame, for example, a square frame with a side length of 50 pixels. The size of the sampling frame should not be too large to avoid collecting white pixels in non-vitreous areas and affecting the identification of the vitreous area.

[0082] The specific process is: the sampling frame traverses the effective area of ​​the current retinal image from right to left and from top to bottom in sequence. At each sampling, the sampling frame selects a sampling area in the effective area of ​​the current retinal image, and then calculates the average brightness value of all pixels in the sampling area. If the average brightness value is within the preset brightness range, the sampling area is determined to be a vitreous area; if the average brightness value of the sampling area is outside the preset brightness range, it is determined that the sampling area is not a vitreous area, and then moves to the next position for sampling until a vitreous area is determined. The preset brightness range is, for example, 30-40.

[0083] It should be noted that the vitreous region determined in this embodiment is only a part of the vitreous region, not the complete vitreous region.

[0084] Then, the edema brightness threshold of the current retinal image can be determined based on the brightness value of the pixels in the vitreous region of the current retinal image. Please refer to Figure 4 , step 102 specifically includes the following sub-steps:

[0085] Sub-step 1021, obtaining an initial brightness threshold of the brightness values ​​of pixels greater than a preset percentage of all pixels in the vitreous region, wherein the preset percentage is greater than 50%.

[0086] Sub-step 1022, searching for target pixels whose brightness values ​​are within a set brightness value range in the vitreous area, and based on the target brightness value of each target pixel, counting the number of target pixels of each target brightness value; wherein the set brightness value range is a brightness value range with the initial brightness threshold as the midpoint.

[0087] Sub-step 1023, searching among all target brightness values ​​for a target brightness value having the smallest difference in the number of target pixels between the target brightness value and the initial brightness threshold as the minimum change brightness threshold.

[0088] Sub-step 1024, determining an edema brightness threshold based on the initial brightness threshold and the minimum change brightness threshold.

[0089] Specifically, in the aforementioned step 102, a vitreous body region is determined, and all pixels in the vitreous body region can be arranged in ascending order of brightness, and the brightness value that can cover a preset percentage of pixels in the vitreous body region is set as the initial brightness threshold; if the preset percentage is 95%, for example, the determined initial brightness threshold is greater than the brightness values ​​of 95% of the pixels in the vitreous body region.

[0090] A set brightness value range is obtained with the initial brightness threshold as the center, for example, a brightness range of plus or minus 10 brightness of the initial brightness threshold is used as the set brightness value range; for example, if the initial brightness threshold is 35, the set brightness value range is (25, 45).

[0091] Then, among all the pixels in the vitreous area, pixels whose brightness values ​​are within the set brightness value range are searched and recorded as target pixels. Each target pixel has a corresponding brightness value recorded as the target brightness value, and then the number of target pixels corresponding to each target brightness value is counted; the initial brightness threshold also has its corresponding number of target pixels, and then the difference between the number of target pixels of the initial brightness threshold and the number of target pixels of each target brightness value is calculated respectively, and the target brightness value corresponding to the minimum difference is selected as the minimum change brightness threshold.

[0092] Then, the product of the mean value between the initial brightness threshold and the minimum change brightness threshold and the preset correction coefficient is obtained as the edema brightness threshold. The preset correction coefficient is greater than 0.5 to reduce excessive recognition of subsequent edema areas; for example, the preset correction coefficient is 0.8, the edema brightness threshold P = 0.8 × (L1 + L2) / 2, L1 represents the initial brightness threshold, L2 represents the minimum change brightness threshold, and the edema brightness threshold obtained by setting the preset correction coefficient to 0.8 can improve the accuracy of the subsequently identified edema areas to a certain extent.

[0093] Step 103, based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, determine the edema area in the current retinal image, and the brightness value of the pixel in the edema area is less than the edema brightness threshold.

[0094] Specifically, based on the edema brightness threshold, the position of the characteristic pixel whose brightness value is less than the edema brightness threshold is searched in the current retinal image, and then the area composed of the characteristic pixels found is used as the edema area; that is, based on the edema brightness threshold as the boundary, the current retinal image is binary judged to distinguish the characteristic pixels of macular edema and the noise pixels of non-macular edema in the current retinal image, so as to realize the feature extraction of the edema area of ​​the current retinal image. The feature extraction here can be understood as the determination of pixels belonging to the edema area; for example, the pixels in the current retinal image are binarized with the edema brightness threshold as the boundary, and the pixels with brightness values ​​greater than the edema brightness threshold are the background area and are assigned a value of 0; the pixels with brightness values ​​less than or equal to the edema brightness threshold are the possible edema area and are assigned a value of 1; and a binary image of the current retinal image can be obtained. Exemplarily, when the feature extraction of the edema area of ​​the current retinal image is performed, it can also be performed only for the effective area in the current retinal image.

[0095] The pixel positions mentioned in this embodiment and subsequent embodiments are the coordinates of the pixels in the image. A coordinate system is constructed with the pixel in the lower left corner of the image as the origin, the bottom row of pixels as the X-axis, and the leftmost column of pixels as the Y-axis. The positions of all pixels in the image can be coordinate-based.

[0096] In one example, based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, the edema area in the current retinal image is determined, including: in the closed area composed of pixels with brightness values ​​less than the edema brightness threshold in the current retinal image, a closed area whose area meets the preset conditions is selected as the edema area; the preset condition is that the area of ​​the closed area is less than the product of the effective area area in the current retinal image and the set threshold and is greater than the preset pixel number area, the set threshold is greater than 0 and less than 0.5, for example, 0.2, and the preset pixel number area is the area of ​​a certain number of pixels, such as the area of ​​two pixels; thereby, the misidentification of the vitreous area can be reduced and noise reduction can be achieved. Please refer to Figure 5 , the red pixels marked in the current retinal image are the identified edema areas.

[0097] In one example, after step 103, the method further includes:

[0098] Step 104 : For each edema region, the area of ​​the edema region is determined based on the pixel spacing information of the current retinal image and the number of pixels in the edema region.

[0099] Specifically, after determining the edema area included in the current retinal image, the number of pixels in each edema area is counted respectively; based on the pixel spacing information of the current retinal image, the area of ​​a single pixel in the current retinal image can be calculated; then the number of pixels in each edema area is multiplied by the area of ​​a single pixel to obtain the area of ​​each edema area. Thus, the areas of all edema areas are added together to obtain the total area of ​​edema areas in the current retinal image.

[0100] Among them, the pixel spacing information of the current retinal image can be extracted from the DICOM file of the current retinal image. The pixel spacing information represents the length of each pixel in the actual space, including the two pixel spacings (x1, y1) of the x-axis and the y-axis. x1 can be equal to y1, indicating a square pixel, and x1 can also be equal to y1, indicating a rectangular pixel; the pixel spacing is, for example, (0.5, 0.5) mm, and a certain edema area includes K pixels, then the area of ​​the edema area is 0.25K, in square millimeters.

[0101] Step 105 , based on the area of ​​each edema region in the current retinal image and the image distance information between the current retinal image and the adjacent retinal images, obtain the volume of each edema region in the current retinal image.

[0102] Specifically, image spacing information can be extracted from the DICOM file of the current retinal image, and the image spacing information represents the distance between the current retinal image and the adjacent previous retinal image slice; the area of ​​each edema region on the current retinal image has been obtained in step 104, and for each edema region, the area of ​​the edema region is integrated using the above-mentioned image spacing information to obtain the volume of the edema region, and the integration method is, for example, the Simpson integration method; repeating the above process can obtain the volume of all edema regions on the current retinal image relative to the previous retinal image.

[0103] By repeating the above process, the volume of the edema area of ​​the second to the last retinal image relative to the corresponding previous retinal image can be obtained. The volume of the macular edema of the current user can be obtained by adding up the volumes of all the edema areas; that is, the area of ​​each edema area can be integrated separately with the help of the image spacing information, so that the calculated edema volume has a higher accuracy.

[0104] Furthermore, after determining the edema areas in all retinal images, etdrs partitions are first performed on the current retinal image to obtain the edema areas within the etdrs partitions. For example, multiple circles are drawn with the central fovea of ​​the macula as the center to further segment the edema area. Specifically, the central fovea of ​​the macula is taken as the center with diameters of 1mm, 3mm, and 6mm, and then the circle with a diameter of 6mm is divided into four equal parts. Thus, the edema area can be divided into fan-shaped ring areas with diameters of 1mm, 1-3mm, and 3-6mm. Furthermore, the quartering line can be extended to divide the area outside the 6mm diameter circle into four areas, and the edema in these areas is also included in the subsequent edema area calculation. Figure 6 For example, area 0 is a circle with a diameter of 1 mm, areas 0 to 4 form a circle with a diameter of 3 mm, areas 0 to 8 form a circle with a diameter of 6 mm, and the rectangular frame is the extended area of ​​the circle with a diameter of 6 mm, which is a 6 mm × 6 mm square. Areas 9-12 in the square will also be included in the subsequent edema area calculation; in some examples, the edema area mark outside the circle with a diameter of 6 mm can also be deleted.

[0105] When the area and volume of the edema area are subsequently calculated, the edema area is calculated for the edema area in the etdrs partition (circular area) and the extended area (6mm×6mm square); if the edema area mark outside the circle is deleted, the area and volume can be calculated only for the edema area in the etdrs partition (circular area). The etdrs partition is a macular partition method defined by the Institute for Early Treatment of Diabetic Retinopathy. The retinal image is partitioned by etdrs, which can obtain edema areas with certain clinical medical significance. Therefore, when calculating the area and volume of the edema area in the etdrs partition, the calculated edema area and volume can have more prominent clinical indication significance.

[0106] In steps 104 to 105, after all edema areas of the current retinal image are determined, the area and volume of the edema area in the current retinal image can be calculated in combination with the pixel spacing information and image spacing information of the current retinal image, that is, the characteristics of the edema area in the retinal image are quantitatively characterized, and quantitative analysis of the edema area is achieved, which can provide accurate quantitative analysis results for reference.

[0107] After calculating the edema area and edema area volume of the current retinal image, the annotation information of the current retinal image can be further enriched, and an OCT image group with the final edema area annotated on the current retinal image can be output, including the shooting date, serial number, patient ID, equipment serial number, shooting date, edema area array, edema volume file, the original OCT image of the current retinal image, and the OCT image with the final edema area marked.

[0108] In this embodiment, the edema brightness threshold of the retinal image can be determined based on the brightness values ​​of the pixels in the vitreous area of ​​the retinal image, and then the pixels in the retinal image can be binary judged as edema and noise based on the edema brightness threshold, so that non-edema noise in the retinal image can be removed, and the feature extraction of macular edema in the retinal image can be achieved; that is, based on the pixel-level information of the retinal image, the feature extraction of macular edema can be automatically performed, and the macular edema area of ​​the retinal image can be obtained by using only a computer through automatic encoding and compilation technology. Compared with macular edema recognition through machine learning, it does not require training through data or high-performance hardware, and is simpler and easier to use; it achieves a higher signal-to-noise ratio, has the characteristics of high accuracy, fast recognition speed and low hardware requirements, and is suitable for a variety of macular edema feature extraction scenarios.

[0109] The second embodiment of the present invention relates to a method for acquiring edema features of macular degeneration in a retinal image. Compared with the first embodiment, this embodiment adds correction of the edema area in the current retinal image.

[0110] like Figure 7 As shown, it is a specific flow chart of the method for acquiring macular lesion edema features in retinal images of this embodiment.

[0111] Step 201, obtaining a current retinal image to be identified from a plurality of retinal images, is substantially the same as step 101 in the first embodiment and will not be described in detail herein.

[0112] Step 202 , performing various preprocessing on the current retinal image to obtain a plurality of reference retinal images after various preprocessing.

[0113] Specifically, the current retinal image is preprocessed in different ways, including but not limited to: non-local mean processing, automatic contrast adjustment processing, sharpening processing, and Otsu method processing.

[0114] Take the current retinal image as an example after the above four preprocessings:

[0115] Method 1: Non-local mean processing, such as the fast non-local mean denoising method, which takes into account the self-similarity of the image, makes full use of the redundant information in the image, and can maintain the detailed features of the image to the greatest extent while denoising; the non-local mean processing will lose the edge information of the image, and the image obtained by the non-local mean processing will generally be sharpened to enhance the edge information of the image. The sharpening method is, for example, image convolution operation, such as using filter2D convolution, using a 3*3 matrix operator (for example, [0, -1, 0], [-1, 5, -1], [0, -1, 0]) to convolve the image. Figure 8 As shown, Figure 8a is the original image of the current retinal image, Figure 8 b is the reference retinal image obtained by non-local mean processing of the current retinal image, Figure 8 c is Figure 8 The reference retinal image in b is sharpened.

[0116] Method 2: Automatic contrast adjustment processing, which automatically enhances the contrast of the current retinal image and improves the contrast between the bright and dark areas in the image, thereby making the brightness difference between the edema area and the surrounding medium in the current retinal image more obvious, and obtaining the corresponding reference retinal image. Fig. 9 As shown, Fig. 9 a is the original image of the current retinal image, Fig. 9 b is the reference retinal image obtained by contrast enhancement of the current retinal image.

[0117] Method 3: Sharpening processing, directly sharpening the current retinal image to enhance the edge information between the bright and dark parts of the image. For details, please refer to the sharpening processing of method 1, which will not be repeated here. Fig.10 As shown, Fig.10 a is the original image of the current retinal image, Fig.10 b is the reference retinal image obtained by sharpening the current retinal image.

[0118] Method 4: Otsu processing, the current retinal image is processed by Otsu method, using Otsu threshold (Otsu), according to the distribution law of pixels of different brightness, the darker peak is defined as the background, the brighter peak is defined as the foreground, and the brightness that maximizes the inter-class variance of the foreground and background images is selected as the brightness threshold. Otsu processing is not affected by brightness and contrast, and can effectively highlight the foreground information in the current retinal image. Fig.11 As shown, Fig.11 a is the original image of the current retinal image, Fig.11 b is the reference retinal image obtained by processing the current retinal image with the Otsu method.

[0119] Step 203 : determining the edema area in each reference retinal image based on the brightness value of each pixel in each reference retinal image and the edema brightness threshold of each reference retinal image.

[0120] Specifically, the edema area in each reference retinal image is determined, wherein the specific method of determining the edema brightness threshold of the reference retinal image and the edema area in the reference retinal image is similar to the method of determining the edema area of ​​the current retinal image in steps 102 and 103 of the first embodiment, and will not be repeated here.

[0121] Step 204, based on the brightness values ​​of the pixels in the vitreous region of the current retinal image, determines the edema brightness threshold of the current retinal image. This is substantially the same as step 102 in the first embodiment and will not be described in detail herein.

[0122] Step 205, based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, determines the edema area in the current retinal image. This is substantially the same as step 103 in the first embodiment and will not be described in detail here.

[0123] Step 206, determining the effective edema area in the current retinal image based on the edema area of ​​each reference retinal image and the edema area of ​​the current retinal image and preset weight parameters, wherein the preset weight parameters include: weights of each reference retinal image and the current retinal image.

[0124] After the above process, the current retinal image has undergone the above four preprocessings to obtain four reference retinal images. After the four reference retinal images and the current retinal image are identified through edema area, five images containing possible edema areas are obtained. The pixels in the five images are binarized respectively, and the value of the pixel in the edema area is 1, and the value of the pixel in the background area is 0; then the value of the pixel at each position in the five images is multiplied by the weight corresponding to each image and a sum is calculated to obtain the score of the pixel at each position; for the pixel at each position, if the score of the pixel is greater than a preset score (for example, 70), it is determined that the pixel belongs to a valid edema area; otherwise, it is determined that the pixel does not belong to a valid edema area;

[0125] The four reference retinal images and the current retinal image all have corresponding weights, such as method 1: the weight of the reference retinal image obtained by non-local mean processing is 40, method 2: the weight of the reference retinal image obtained by automatic contrast adjustment processing is 20, method 3: the weight of the reference retinal image obtained by sharpening processing is 50, method 4: the weight of the reference retinal image obtained by Otsu method processing is 20; the weight of the current retinal image is 40. For example, the values ​​of the pixels at a certain position on the five images are 0, 1, 1, 0, 1 respectively, then the score of the pixel = (40×0) + (20×1) + (50×1) + (20×0) + (40×1) = 110, 110 is greater than the preset score of 70, and it is determined that the pixel belongs to the effective edema area.

[0126] Based on the above process, it is possible to determine whether each pixel in the current retinal image belongs to a valid edema area, and thus the valid edema area in the current retinal image can be obtained.

[0127] Step 207 : For each edema region, the area of ​​the edema region is determined based on the pixel spacing information of the current retinal image and the number of pixels in the edema region.

[0128] Step 208: obtaining the volume of each edema region in the current retinal image based on the area of ​​each edema region in the current retinal image and the image spacing information of the current retinal image.

[0129] Step 207 and step 208 are similar to step 104 and step 105 in the first embodiment and are not described in detail here. The main difference is that step 207 and step 208 are aimed at the effective edema area determined in step 206.

[0130] Subsequently, the reference retinal images obtained by the four preprocessing processes may be saved, and the identified edema areas, sampling areas, etc. may be marked therein; similarly, the current retinal image with the final edema area marked may be saved.

[0131] In this embodiment, the current retinal image is subjected to multiple preprocessing methods, and different preprocessing methods can highlight different image features. Subsequently, the edema area in the current retinal image is adjusted in combination with the edema area in the reference retinal image obtained after the preprocessing of the current retinal image to obtain the final effective edema area, thereby achieving a better denoising effect and further improving the accuracy of edema area feature extraction, that is, further improving the segmentation accuracy of the edema area in the retinal image.

[0132] The third embodiment of the present invention relates to a method for acquiring macular edema features in retinal images. Compared with the first embodiment, this embodiment adds a denoising process based on RPE layer recognition.

[0133] like Fig.12 As shown, it is a specific flow chart of the method for acquiring macular lesion edema features in retinal images of this embodiment.

[0134] Step 301, obtaining a current retinal image to be identified from a plurality of retinal images, which is substantially the same as step 101 in the first embodiment and will not be described in detail herein.

[0135] Step 302, based on the brightness values ​​of the pixels in the vitreous region of the current retinal image, determines the edema brightness threshold of the current retinal image. This is substantially the same as step 102 in the first embodiment and will not be described in detail herein.

[0136] Step 303: Identify the RPE layer of the current retinal image based on the brightness value of each pixel in the current retinal image.

[0137] Specifically, the effective area of ​​the current retinal image is first obtained, and the relevant content in the first embodiment may be referred to for details. Then, the target pixel with the highest brightness is searched by column within the effective area of ​​the current retinal image. For the target pixel in the current column within the searched effective area of ​​the current retinal image, if the longitudinal position difference between the target pixel in the current column and the target pixel in the previous column adjacent to the current column is greater than a first difference threshold, the longitudinal position of the target pixel in the current column is determined based on the longitudinal position of the target pixel in the previous column adjacent to the current column.

[0138] In one example, starting from any middle column of the effective area of ​​the current retinal image, the target pixel with the highest brightness in each column is searched to the left and right respectively. For example, the middle column is selected from multiple columns in the effective area of ​​the current retinal image (if the total number of columns is an even number, any one of the two middle columns is selected), and identification is performed column by column starting from the middle column to the right. For the current column, the target pixel with the highest brightness in the current column is first found, and then the Y-axis position of the target pixel is compared with the Y-axis position of the target pixel in the previous column. If the difference between the Y-axis positions is less than the first difference threshold, it is determined that the target pixel in the current column belongs to the RPE layer; if the difference between the Y-axis positions is greater than the first difference threshold (for example, 20 pixels), it means that there may be other highlight positions on the image that cause interference, and it is determined that the target pixel in the current column does not belong to the RPE layer, and the Y-axis position of the target pixel in the previous column is used as the Y-axis position of the target pixel in the current column, and the X-axis position of the target pixel in the current column remains unchanged. Then, the pixels are identified column by column from the middle column to the left of the effective area of ​​the current retinal image, and the above process is repeated to determine the target pixels belonging to the RPE layer in the columns to the left of the middle column. Starting pixel identification from the middle column of the effective area of ​​the current retinal image can avoid the influence of image edge blur on the accuracy of RPE layer identification.

[0139] From the above, the target pixels belonging to the RPE layer in each column of the effective area of ​​the current retinal image can be determined. It should be noted that in the recognition process, if there are multiple columns (the specific number can be determined based on the total number of columns in the effective area, such as one quarter of the total number of columns) in the effective area of ​​the current retinal image, and the difference between the longitudinal axis position of the target pixel in the previous column is greater than the first difference threshold, it is necessary to consider whether there is a problem with the longitudinal axis position of the target pixel in the column to be recognized; or there is too much highlight interference in the effective area of ​​the current retinal image.

[0140] Then, the RPE layer of the current retinal image is obtained by combining the target pixels in each column in the effective area of ​​the current retinal image. The positions of all target pixels in the RPE layer are then recorded to form a corresponding coordinate array.

[0141] In addition, if the current retinal image is associated with multiple designated retinal images, the position of the RPE layer of the current retinal image can also be corrected based on the position of the RPE layer of these multiple designated retinal images; the specific process is as follows: the multiple designated retinal images associated with the current retinal image refer to multiple OCT images (OCT images in the same group) acquired continuously from the same patient during one OCT image acquisition, and all of these multiple OCT images except the current retinal image can be used as designated retinal images; the positions of the RPE layers of these multiple OCT images should be similar.

[0142] For these multiple specified retinal images, based on the above-mentioned similar process, the position of the RPE layer in the effective area of ​​each specified retinal image is determined respectively; then based on the average height of the pixels contained in the RPE layer of the multiple specified retinal images, the pixel reference height of the RPE layer is determined; that is, for the RPE layer of each specified retinal image, the mean of the longitudinal position coordinates of all pixels of the RPE layer of the specified retinal image is calculated as the pixel average height of the RPE layer of the specified retinal image; thus, the pixel average height of the RPE layer of all specified retinal images can be calculated. Then the mean of the pixel average height of the RPE layer of all specified retinal images is calculated as the pixel reference height of the RPE layer; however, it is not limited to this, and the median of the pixel average height of the RPE layer of all specified retinal images can also be selected as the pixel reference height of the RPE layer.

[0143] After determining the pixel reference height of the RPE layer, the pixel average height of the pixels included in the RPE layer of the current retinal image is compared with the pixel reference height. If the difference between the pixel average height of the pixels included in the RPE layer of the current retinal image and the pixel reference height is greater than the second difference threshold, it means that the position of the RPE layer of the effective area of ​​the determined current retinal image is inaccurate, and the RPE layer of the current retinal image is determined based on the longitudinal axis position of the pixels included in the RPE layer of a specified retinal image adjacent to the current retinal image; that is, the longitudinal axis position of the pixels included in the RPE layer of the specified retinal image adjacent to the current retinal image is used as the longitudinal axis position of the pixels of the RPE layer of the current retinal image, thereby re-determining the RPE layer in the current retinal image. Wherein, the specified retinal image adjacent to the current retinal image can be a specified retinal image that is closest to the acquisition time of the current retinal image.

[0144] After the above process, after the RPE layer of the current retinal image is determined, the RPE layer can be marked on the current retinal image. It is impossible for the area below the RPE layer to have an edema area. Therefore, based on the position of the RPE layer in the retinal image, the area below the RPE layer can be avoided from being misidentified as a macular edema area.

[0145] Furthermore, after step 303, the following steps are further included:

[0146] Step 304, obtaining multiple specified retinal images associated with the current retinal image, and determining the pixel reference height of the RPE layer based on the average height of the pixels included in the RPE layer of the multiple specified retinal images; if the difference between the pixel average height of the pixels included in the RPE layer of the current retinal image and the pixel reference height is greater than a second difference threshold, then determining the RPE layer of the current retinal image based on the longitudinal axis position of the pixels included in the RPE layer of a specified retinal image adjacent to the current retinal image.

[0147] That is, the RPE layers of multiple specified retinal images (e.g., 5 images on the left and right of the current retinal image) adjacent to the current retinal image among the multiple retinal images obtained by slicing a three-dimensional stereoscopic image of the user are respectively identified to obtain the RPE layers of the multiple specified retinal images, and then the pixel average heights of the pixels included in the RPE layer in each of the specified retinal images are calculated to obtain the pixel average height of the RPE layer in the multiple specified retinal images, and the pixel reference height is determined based on the multiple pixel average heights, for example, the median of the pixel average heights of the RPE layer in the multiple specified retinal images is selected as the pixel reference height, or the mean is calculated as the pixel reference height. reference height; then the average pixel height of the pixels in the RPE layer of the current retinal image is compared with the pixel reference height. If the difference between the two is greater than a second difference threshold (for example, 20 pixels), it is determined that an error has occurred in the recognition of the RPE layer of the current retinal image. If an error has occurred in the recognition of the RPE layer of the current retinal image, the coordinates of each pixel in the RPE layer in the previous or next specified retinal image adjacent to the current retinal image are used as the coordinates of the RPE layer in the current retinal image, thereby re-determining an RPE layer in the current retinal image; if the difference between the two is less than the second difference threshold, it is determined that there is no error in the recognition of the RPE layer of the current retinal image.

[0148] The height of the pixel is the Y-axis coordinate of the pixel. The coordinate system takes the pixel at the lower left corner of the retinal image as the origin, the Y-axis is upward from the origin, and the X-axis is to the right of the origin.

[0149] Step 305, based on the RPE layer of the current retinal image, the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, determine the edema area in the current retinal image, and all pixels in the edema area are located above the RPE layer.

[0150] It is substantially the same as step 103 in the first embodiment and will not be described in detail here. The main difference is that the edema area feature extraction of the current retinal image can be performed for denoising based on the position of the RPE layer in the current retinal image. The specific method is as follows:

[0151] Method 1: After determining the position of the RPE layer in the current retinal image, adjust the part below the RPE layer in the current retinal image to black, because the area below the RPE layer is close to the macular edema area in brightness and shape. Therefore, by blackening, the area below the RPE layer can be prevented from being misidentified as the macular edema area. Fig.13 ,in Fig.13 a is the original forward retinal image; Fig.13 b is the current retinal image with the part below the RPE layer adjusted to black; at this time, the edema area in the current retinal image is determined based on the brightness value of each pixel in the current retinal image with the part below the RPE layer adjusted to black and the edema brightness threshold of the current retinal image; that is, when determining the edema area in the current retinal image, the current retinal image with the part below the RPE layer adjusted to black is targeted. Please refer to Fig.14 , the part below the RPE layer is adjusted to black, and the red pixels in the current retinal image are the identified edema areas. It can be seen that the noise below the RPE layer is eliminated.

[0152] Method 2 is to first extract the features of the edema area in the current retinal image to determine the edema area in the current retinal image, and then remove the edema area below the RPE layer based on the position of the RPE layer in the current retinal image and the position of each pixel in the determined edema area, thereby avoiding the area below the RPE layer from being misidentified as a macular edema area.

[0153] Step 306, for each edema region, based on the pixel spacing information of the current retinal image and the number of pixels in the edema region, the area of ​​the edema region is determined. This is substantially the same as step 104 in the first embodiment and will not be described in detail herein.

[0154] Step 307, based on the area of ​​each edema region in the current retinal image and the image spacing information between the current retinal image and the adjacent retinal image, obtain the volume of each edema region in the current retinal image. This is substantially the same as step 105 in the first embodiment and will not be described in detail here.

[0155] It should be noted that the present embodiment can also be used as an improvement based on the second embodiment. Specifically, after obtaining a plurality of reference retinal images after various preprocessing, the RPE layer of each reference retinal image can be determined based on the method of the present embodiment, and based on the position of the RPE layer of each reference retinal image, the edema area in each determined reference retinal image is denoised to avoid the area below the RPE layer being misidentified as a macular edema area.

[0156] The fourth embodiment of the present invention relates to an electronic device, which may be a common computer device, such as a desktop host, a laptop computer, etc., or a mobile phone.

[0157] The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for acquiring macular lesion edema characteristics in retinal images of any one of the first to third embodiments.

[0158] Since the first embodiment corresponds to this embodiment, this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and the technical effects that can be achieved in the first embodiment can also be achieved in this embodiment. In order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the first embodiment.

[0159] It should be noted that the memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. Similarly, the processor may also be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0160] A third embodiment of the present invention relates to a computer-readable storage medium, which is a non-volatile storage medium or a non-transitory storage medium, on which a computer program is stored, and is characterized in that when the computer program is executed by a processor, the method for acquiring macular lesion edema characteristics in retinal images of any one of the first to third embodiments is executed.

[0161] Since the first embodiment corresponds to this embodiment, this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and the technical effects that can be achieved in the first embodiment can also be achieved in this embodiment. In order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the first embodiment.

[0162] Preferred embodiments of the present invention have been described above in detail, but it should be understood that aspects of the embodiments can be modified, if necessary, to employ aspects, features and concepts of the various patents, applications and publications to provide further embodiments.

[0163] These and other changes can be made to the embodiments in light of the above detailed description.In general, in the claims, the terms used should not be considered limited to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments along with the full scope of equivalents to which these claims are entitled.

Claims

1. A method for acquiring macular edema features in retinal images, characterized in that: include: Obtaining a current retinal image to be identified from multiple retinal images; Determining an edema brightness threshold of the current retinal image based on brightness values ​​of pixels in the vitreous region of the current retinal image; Based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, an edema area in the current retinal image is determined, and the brightness value of the pixels in the edema area is less than the edema brightness threshold.

2. The method for acquiring macular edema features in retinal images according to claim 1, characterized in that: The method further comprises: Performing a plurality of preprocessing operations on the current retinal image to obtain a plurality of reference retinal images after the preprocessing operations; Determining an edema area in each of the reference retinal images based on a brightness value of each pixel in each of the reference retinal images and an edema brightness threshold of each of the reference retinal images; Based on the edema area of ​​each of the reference retinal images and the edema area of ​​the current retinal image and preset weight parameters, the effective edema area in the current retinal image is determined, and the preset weight parameters include: the weights of each of the reference retinal images and the current retinal image.

3. The method for acquiring macular edema features in retinal images according to claim 1, characterized in that: Before determining the edema area in the current retinal image based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, the method further includes: Identifying the RPE layer of the current retinal image based on the brightness value of each pixel in the current retinal image; Determining an edema area in the current retinal image based on a brightness value of each pixel in the current retinal image and an edema brightness threshold of the current retinal image includes: Based on the RPE layer of the current retinal image, the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, the edema area in the current retinal image is determined, and the pixels in the edema area are all located above the RPE layer.

4. The method for acquiring macular edema features in retinal images according to claim 1, characterized in that: Determining an edema brightness threshold of the current retinal image based on brightness values ​​of pixels in the vitreous region of the current retinal image comprises: Obtaining an initial brightness threshold value of a brightness value of pixels greater than a preset percentage of all pixels in the vitreous region, wherein the preset percentage is greater than 50%; Searching for target pixels whose brightness values ​​are within a set brightness value range in the vitreous region, and counting the number of target pixels of each target brightness value based on the target brightness value of each target pixel; wherein the set brightness value range is a brightness value range with the initial brightness threshold as the midpoint; Finding a target brightness value with the smallest target pixel number difference from the initial brightness threshold among all the target brightness values ​​as the minimum change brightness threshold; The edema brightness threshold is determined based on the initial brightness threshold and the minimum change brightness threshold.

5. The method for acquiring macular lesion edema features in retinal images according to claim 4, characterized in that: Determining the edema brightness threshold based on the initial brightness threshold and the minimum change brightness threshold includes: The product of the average value between the initial brightness threshold and the minimum change brightness threshold and a preset correction coefficient is obtained as the edema brightness threshold, and the preset correction coefficient is greater than 0.

5.

6. The method for acquiring macular edema features in retinal images according to claim 1, characterized in that: Determining an edema area in the current retinal image based on a brightness value of each pixel in the current retinal image and an edema brightness threshold of the current retinal image includes: In the closed area formed by pixels whose brightness values ​​in the current retinal image are less than the edema brightness threshold, a closed area whose area meets a preset condition is selected as an edema area; The preset condition is that the area of ​​the closed area is smaller than the product of the area of ​​the effective area in the current retinal image and a set threshold and is larger than the area of ​​a preset number of pixels, and the set threshold is greater than 0 and less than 0.

5.

7. The method for acquiring macular lesion edema features in retinal images according to claim 1, characterized in that: After determining the edema area in the current retinal image based on the brightness value of each pixel in the current retinal image and the edema brightness threshold of the current retinal image, the method further includes: For each of the edema regions, the area of ​​the edema region is determined based on the pixel spacing information of the current retinal image and the number of pixels in the edema region.

8. The method for acquiring macular edema features in retinal images according to claim 7, characterized in that: After determining the area of ​​each edema region based on the pixel spacing information of the current retinal image and the number of pixels in the edema region, the method further includes: Based on the area of ​​each edema region in the current retinal image and the image distance information between the current retinal image and an adjacent retinal image, the volume of each edema region in the current retinal image is obtained.

9. The method for acquiring macular edema features in retinal images according to claim 7 or 8, characterized in that: Before determining the area of ​​each edema region based on the pixel spacing information of the current retinal image and the number of pixels in the edema region, the method further includes: Performing etdrs partitioning on the current retinal image to obtain the edema area within the etdrs partitioning; For each of the edema regions, determining the area of ​​the edema region based on the pixel spacing information of the current retinal image and the number of pixels in the edema region comprises: For each of the edema regions within the etdrs partition, the area of ​​the edema region is determined based on the pixel spacing information of the current retinal image and the number of pixels within the edema region.

10. The method for acquiring macular edema features in retinal images according to claim 3, characterized in that: Identifying the RPE layer of the current retinal image based on the brightness value of each pixel in the current retinal image, comprising: Searching for a designated pixel with the highest brightness in a valid area of ​​the current retinal image by column, and for a designated pixel in a current column in the valid area of ​​the current retinal image that has been searched, if a longitudinal position difference between the designated pixel in the current column and the designated pixel in a previous column adjacent to the current column is greater than a first difference threshold, determining a longitudinal position of the designated pixel in the current column based on the longitudinal position of the designated pixel in the previous column adjacent to the current column; The RPE layer of the current retinal image is obtained by combining the designated pixels in each column within the effective area of ​​the current retinal image.

11. The method for acquiring macular edema features in retinal images according to claim 10, characterized in that: Searching for target pixels with the highest brightness in columns within the effective area of ​​the current retinal image includes: Starting from any middle column in the effective area of ​​the current retinal image, search leftward and rightward respectively for the target pixel with the highest brightness in each column.

12. The method for acquiring macular edema features in retinal images according to claim 10, characterized in that: After obtaining the RPE layer of the current retinal image by combining the designated pixels in each column within the effective area of ​​the current retinal image, the method further includes: Acquire a plurality of designated retinal images associated with the current retinal image, and determine a pixel reference height of the RPE layer based on an average height of pixels included in the RPE layer of the plurality of designated retinal images; If the difference between the average pixel height of the pixels included in the RPE layer of the current retinal image and the pixel reference height is greater than a second difference threshold, the RPE layer of the current retinal image is determined based on the longitudinal position of the pixels included in the RPE layer of the specified retinal image adjacent to the current retinal image.

13. The method for acquiring macular edema features in retinal images according to claim 1 or 4, characterized in that: The vitreous area of ​​the current retinal image is determined in the following manner: The effective area of ​​the current retinal image is sampled in sequence using a sampling frame of a set size, and if the average brightness value of the pixels in the sampling area currently selected by the sampling frame is within a preset brightness range, it is determined that the sampling area currently selected by the sampling frame is the vitreous area.

14. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for acquiring macular lesion edema features in retinal images as described in any one of claims 1 to 13.

15. A computer-readable storage medium, wherein the computer-readable storage medium is a non-volatile storage medium or a non-transient storage medium, and a computer program is stored thereon, wherein: When the computer program is executed by a processor, the method for acquiring macular edema features in a retinal image according to any one of claims 1 to 13 is performed.

Images