KR102742549B1 - Electronic device for Optic disc localization in fundus images and method thereof - Google Patents

Abstract

In one embodiment of the present invention, an electronic device for identifying a position of an optic disc in a fundus image comprises a processor configured to perform motion blur on an input fundus image at different angles to obtain a directional blur image, identify a plurality of candidate optic discs using an extended maximum transform technique from the directional blur image, perform radial blur on the directional blur image with respect to the center of each candidate optic disc as a reference, and, when the radial blur image is normal, identify a first candidate optic disc having the largest maximum intensity value among the plurality of candidate optic discs as the optic disc, and, when the radial blur image is abnormal, identify a second candidate optic disc having the widest intensity distribution when the width of the maximum intensity value is 80% of the maximum intensity value as the optic disc.

Description

{Electronic device for Optic disc localization in fundus images and method thereof}

The present invention relates to an electronic device and method for identifying the position of the optic nerve head in a fundus image.

One of the key steps in the analysis of fundus images (the inner surface of the eye, opposite the lens) for diagnosing eye diseases and localizing various anatomical structures is to determine the location of the optic disc.

The optic disc is the passageway through which the optic nerve and blood vessels enter the retina and usually appears as a bright yellow or white area on color fundus images.

However, the optic nerve head can easily be confused with other anatomical structures (hereinafter referred to as lesions) such as the fovea, exudates, and retinopathy-related lesions.

To date, optic disc detection and localization have utilized a variety of methods based on a combination of features and techniques including edge, disc shape, blood vessels, color space, and template matching/morphological operations.

For example, one of these methods relies on contrast enhancement, use of the morphological top-hat-transform, and post-filtering (Walter, T., Klein, J.-C., 2001. Segmentation of color fundus images of the human retina: detection of the optic disc and the vascular tree using morphological techniques. In: Proceedings of the International Symposium on Medical Data Analysis. Springer, pp. 282-287.), while another uses Hausdorff-based template matching, edge mapping, and pyramid decomposition (Lalonde, M., Beaulieu, M., Gagnon, L., 2001. Fast and robust optic disc detection using pyramidal decomposition and hausdorff-based template matching. IEEE Trans. Med. Imaging 20, 1193-1200.).

Additionally, a method to identify the location of the optic nerve head using machine learning and deep learning has been proposed (Staal, J., Abr`amoff, M.D., Niemeijer, M., Viergever, M.A., Ginneken, B. Van, 2004. Ridgebased vessel segmentation in color images of the retina. IEEE Trans. Med. Imaging 23, 501-509.).

Machine learning and deep learning-based methods provide reasonable accuracy, but they require a large number of training data and considerable time to learn the model. In addition, it is rather complicated to adjust various parameters including network depth, number of convolutional and sub-sampling layers, learning rate, attenuation, and selecting an appropriate loss function to obtain an effective model.

An object of the present invention is to provide an electronic device and method for identifying the position of the optic nerve head more precisely.

An object of the present invention is to provide an electronic device and method for more accurately identifying the position of the optic nerve head.

An object of the present invention is to provide an electronic device and method for accurately identifying the location of the optic nerve head even when the fundus image is abnormal, such as when a lesion is present.

In one embodiment of the present invention, an electronic device for identifying a position of an optic disc in a fundus image comprises a processor configured to perform motion blur on an input fundus image at different angles to obtain a directional blur image, identify a plurality of candidate optic discs using an extended maximum transform technique from the directional blur image, perform radial blur on the directional blur image with respect to the center of each candidate optic disc as a reference, and, when the radial blur image is normal, identify a first candidate optic disc having the largest maximum intensity value among the plurality of candidate optic discs as the optic disc, and, when the radial blur image is abnormal, identify a second candidate optic disc having the widest intensity distribution when the width of the maximum intensity value is 80% of the maximum intensity value as the optic disc.

The above processor can convert the color space of the fundus image into the CIELAB color space before performing motion blur.

The above processor can obtain a plurality of first blur images by performing motion blur on the fundus image at different angles, and obtain the directional blur image by combining the plurality of first blur images.

The above processor can perform radial blurring with respect to the center of each candidate optic nerve head on the directional blur image to obtain a plurality of second blur images, and combine the plurality of second blur images to obtain the radial blur image.

The above processor can identify an intensity maximum from the intensity distribution of each candidate optic nerve head acquired through the radial blur image.

The above processor can determine whether the radial blur image is normal or abnormal based on whether the ratio of the size of the lesion region to the size of the radial blur image, the size of the lesion region, or the number of lesion regions exceeds a threshold value.

In one embodiment of the present invention, a method for identifying a position of an optic disc in a fundus image performed by an electronic device comprises the steps of: performing motion blur on an input fundus image at different angles to obtain a directional blur image; identifying a plurality of candidate optic discs using an extended maximum transform technique from the directional blur image; performing radial blur on the directional blur image with respect to the center of each candidate optic disc as a reference to obtain a radial blur image; identifying a first candidate optic disc having the largest maximum intensity value among the plurality of candidate optic discs as the optic disc when the radial blur image is normal; and identifying a second candidate optic disc having the widest intensity distribution width when the intensity maximum value is 80% of the intensity maximum value as the optic disc when the radial blur image is abnormal.

The step of obtaining the above directional blur image may include the step of converting the color space of the fundus image into the CIELAB color space before performing motion blur.

The step of obtaining the above directional blur image may include the step of obtaining a plurality of first blur images by performing motion blur on the fundus image at different angles; and the step of obtaining the directional blur image by combining the plurality of first blur images.

The step of obtaining the above radial blur image may include the step of performing radial blur on the directional blur image based on the center of each candidate optic nerve head to obtain a plurality of second blur images; and the step of obtaining the radial blur image by combining the plurality of second blur images.

The method may further include a step of identifying an intensity maximum from an intensity distribution of each candidate optic nerve head acquired through the radial blur image.

The method may further include a step of determining whether the radial blur image is normal or abnormal based on whether a ratio of the size of the lesion region to the size of the radial blur image, the size of the lesion region, or the number of lesion regions exceeds a threshold value.

According to one embodiment of the present invention, the position of the optic nerve head can be effectively identified based on color conversion and blurring.

According to one embodiment of the present invention, the position of the optic nerve head can be precisely identified not only when the fundus image is normal, but also when there is an abnormality such as a lesion.

FIG. 1 is a schematic diagram illustrating an electronic device for identifying a fundus image and the position of the optic nerve head according to one embodiment of the present invention.
FIG. 2 is a block diagram illustrating the configuration of an electronic device according to one embodiment of the present invention.
FIG. 3 is a drawing illustrating an operation flow diagram of an electronic device according to one embodiment of the present invention.
FIG. 4 is a drawing illustrating a motion blur performed according to one embodiment of the present invention.
FIG. 5 is a drawing illustrating an example of identifying the position of the optic nerve head according to one embodiment of the present invention.
FIG. 6 is a diagram illustrating an overall process for identifying the position of the optic nerve head according to one embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The detailed description to be disclosed below together with the accompanying drawings is intended to explain exemplary embodiments of the present invention, and is not intended to represent the only embodiments in which the present invention may be practiced. In the drawings, in order to clearly describe the present invention, parts that are not related to the description may be omitted, and the same reference numerals may be used for the same or similar components throughout the specification.

FIG. 1 is a schematic diagram illustrating an electronic device for identifying a fundus image and the position of the optic nerve head according to one embodiment of the present invention.

According to one embodiment of the present invention, the fundus image (10) means an image obtained by photographing the inner surface of the eyeball.

According to one embodiment of the present invention, the electronic device (100) is a device that identifies the position of the optic disc in an input fundus image (10), and can be implemented as a computer, a server, a smart phone, a tablet PC, a smart pad, a laptop, etc.

The optic disc is usually brighter than other parts of the fundus image, so its location can be identified using high-level grayscale changes. However, since the area containing the lesion is also bright and has good contrast, this method can be applied well to normal images with few lesions.

Therefore, not only does the lesion interfere with accurate detection of the optic disc, but it may itself be a candidate for the optic disc. In addition, the intensity of the optic disc may not be relatively high in fundus images with poor illumination or low contrast.

The present invention proposes an electronic device (100) and method thereof that can more accurately detect the position of the optic nerve head from a fundus image and can be applied even when the fundus image is abnormal, such as when a lesion is present.

Hereinafter, the configuration and operation of an electronic device according to one embodiment of the present invention will be specifically described with reference to the drawings.

FIG. 2 is a block diagram illustrating the configuration of an electronic device according to one embodiment of the present invention.

An electronic device (100) according to one embodiment of the present invention includes an input unit (110), a communication unit (120), a display unit (130), a storage unit (140), and a processor (150).

The input unit (110) generates input data in response to a user input of the electronic device (100). For example, the user input may be a user input for starting the operation of the electronic device (100), a user input for inputting a fundus image, a user input for setting threshold values included in conditions for determining whether a fundus image is normal or abnormal, a maximum intensity value, etc. In addition, if it is a user input necessary for identifying the position of the optic nerve head, it may be applied without limitation.

The input unit (110) includes at least one input means. The input unit (110) may include a keyboard, a key pad, a dome switch, a touch panel, a touch key, a mouse, a menu button, etc.

The communication unit (120) performs communication with external devices such as fundus photographing equipment and servers to receive fundus images, calculation programs necessary to identify the optic nerve head in the fundus images, etc.

To this end, the communication unit (120) can perform wireless communication such as 5G (5th generation communication), LTE-A (long term evolution-advanced), LTE (long term evolution), Wi-Fi (wireless fidelity), Bluetooth, or wired communication such as LAN (local area network), WAN (Wide Area Network), and power line communication.

The display unit (130) displays display data according to the operation of the electronic device (100). The display unit (130) can display a screen displaying a fundus image, a directional blur image, a radioactive blur image, a screen displaying a fundus image with candidate optic discs displayed, a screen displaying a fundus image with identified optic discs displayed, a screen receiving user input, etc.

The display unit (130) includes a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emitting diode (OLED) display, a micro electro mechanical systems (MEMS) display, and an electronic paper display. The display unit (130) may be implemented as a touch screen by being combined with the input unit (110).

The storage (140) stores the operation programs of the electronic device (100). The storage (140) includes non-volatile storage capable of preserving data (information) regardless of whether power is supplied, and volatile memory into which data to be processed by the processor (150) is loaded and which cannot preserving data if power is not supplied. The storage includes flash memory, hard-disc drive (HDD), solid-state drive (SSD), ROM (Read Only Memory), etc., and the memory includes buffer, RAM (Random Access Memory), etc.

The storage unit (140) can store fundus images, directional blur images, radiometric blur images, etc. The storage unit (140) can store computational programs, etc., required in the process of acquiring directional blur images, identifying candidate optic discs, acquiring radiometric blur images, identifying maximum intensity values, and determining whether or not fundus images are normal.

The processor (150) can control at least one other component (e.g., hardware or software component) of the electronic device (100) by executing software such as a program, and can perform various data processing or operations.

A processor (150) according to one embodiment of the present invention performs motion blur at different angles on an input fundus image to obtain a directional blur image, identifies a plurality of candidate optic nerve heads using an extended maximum transform technique from the directional blur image, performs radial blur on the directional blur image with respect to the center of each candidate optic nerve head as a reference, and, when the radial blur image is normal, identifies a first candidate optic nerve head having the highest maximum intensity value among the plurality of candidate optic nerve heads as the optic nerve head, and, when the radial blur image is abnormal, identifies a second candidate optic nerve head having the widest intensity distribution when the intensity maximum value is 80% of the widest intensity value among the plurality of candidate optic nerve heads as the optic nerve head.

Meanwhile, the processor (150) may perform at least some of the data analysis, processing, and result information generation for performing the above operations using at least one of a machine learning, neural network, or deep learning algorithm as a rule-based or artificial intelligence algorithm. Examples of the neural network may include models such as a CNN (Convolutional Neural Network), a DNN (Deep Neural Network), and an RNN (Recurrent Neural Network).

FIG. 3 is a drawing illustrating an operation flow diagram of an electronic device according to one embodiment of the present invention.

According to one embodiment of the present invention, the processor (150) performs motion blur on an input fundus image at different angles to obtain a directional blur image (S10).

According to one embodiment of the present invention, the processor (150) can convert the color space of the fundus image to be suitable for optic nerve head identification and use it. The processor (150) can obtain the fundus image whose color space has been converted before performing motion blur.

Color space is a spatial concept that expresses a color system in three dimensions, and can be expressed in various ways such as RGB (red, green, blue), HSV (hue, saturation, value), YCbCr (luma, blue-luma (B-Y), red-luma (R-Y)), HSL (hue, saturation, brightness), CIE L*a*b* (brightness, red/green, blue/yellow).

In particular, CIE L*a*b* is a color space created by nonlinearly transforming the CIE XYZ color space based on the antagonism theory of human vision, and can provide the best contrast due to the effective suppression of the color and/or intensity of the lesion by the L* component.

The mathematical formula for converting a fundus image in RGB color space to CIELAB color space is as follows.

I(p) is a fundus image with RGB components [I _R , I _G , I _B ](p), and p = (x, y) represents a pixel location. cie(.) is a color space conversion operator from RGB to CIE, and component I _L (p) is used for further processing. Hereinafter, the fundus image is regarded as a color space-converted image.

Meanwhile, studies on blurring and deblurring, which are essential parts of image processing, have confirmed that the optic nerve head contains higher intensity pixels in blurred images.

In addition, the directional blur image generated by the uniform motion blur can provide consistent intensity with noise suppression, so that the candidate optic nerve head can be effectively searched.

Therefore, the processor (150) performs motion blur on the fundus image for better optic disc identification.

Specifically, the processor (150) can perform motion blur on the fundus image at different angles to obtain a plurality of blur images (hereinafter, referred to as a plurality of first blur images), and obtain a directional blur image by combining the plurality of first blur images. A specific process of performing motion blur to obtain a directional blur image is described with reference to FIG. 4.

According to one embodiment of the present invention, the processor (150) identifies a plurality of candidate optic nerve heads using an extended-maxima transform technique from a directional blur image (S20).

Candidate optic discs can be identified by using the extended maximum transform technique, which has properties similar to those of the optic disc. The extended maximum transform is computed based on the local maximum (RE _max ) of the H-max transform using the following equations (2) and (3).

The local maximum (RE _max ) is obtained by calculating the difference between the directional blur image (G(p)) and the H-max (H _max ). The H-max transform suppresses all maxima in the intensity image that have values less than a given threshold t, and is calculated using the following mathematical expression 4.

는 각각 역치 및 확장 재구성을 나타낸다. 여기에서 이진 영상을 얻기 위해 방향성 블러 영상(G(p))에 8개의 연결된 구성 요소가 있는 확장된 최대 변환을 적용한다. 이 바이너리 맵을 사용하면 후보 시신경 유두의 중심과 그 중심 주변의 영역을 방향성 블러 영상에서 쉽게 추출할 수 있다. 방향성 블러 영상에서 식별된 후보 시신경 유두들을 표시한 일 예는 도 5의 (a)에 도시되어 있다. Here t and

represents threshold and dilated reconstruction, respectively. Here, an dilated maximum transform with eight connected components is applied to the directional blur image (G(p)) to obtain a binary image. Using this binary map, the center of the candidate optic disc and the region around the center can be easily extracted from the directional blur image. An example of indicating the candidate optic discs identified in the directional blur image is shown in Fig. 5 (a).

According to one embodiment of the present invention, the processor (150) performs radial blurring based on the center of each candidate optic nerve head on the directional blur image to obtain a radial blur image (S30).

Specifically, the processor (150) may perform radial blurring based on the center of each candidate optic nerve head for the directional blur image to obtain a plurality of blur images (hereinafter, referred to as second blur images), and may combine the plurality of second blur images to obtain a radial blur image. At this time, the center of each candidate optic nerve head may be the center of an area set as the candidate optic nerve head, and the radial blur may be performed using at least one pixel forming the center.

By performing a radial blur on each candidate optic disc, the region can appear brighter against the retinal background, while the textured background is suppressed and smoothed. Therefore, the radial blur can enhance the characteristics of the candidate optic disc region, improve the detectability, and minimize the influence of the overall image variability through smoothing out.

K is the number of candidate optic discs, and D(c _k ) is the image region centered at c _k = (x _k , y _k ) for the kth candidate optic disc (k = {1, 2, ..., K}). Radial blur is applied to each region containing the candidate optic disc.

To generate the k-th blurred image B' _k (c _k ), we first apply a transformation function to a series of images, which are rotated versions of the candidate image D(c _k ) centered at c _k , as in Equation 5. R is a rotation matrix, and θ _i (i = 1, 2, ..., M) is the rotation angle. The maximum rotation number (M) can affect the spatial extent of the intensity smearing.

Finally, all the second blur images are combined to obtain an accumulated radial blur image. The radial blur image can be expressed by mathematical expression 6.

As described above, the fundus image may contain various lesions, and the lesions have high intensity similar to the optic nerve head, which hinders accurate identification of the optic nerve head. Therefore, in the present invention, a method for ultimately identifying the optic nerve head among candidate optic nerve heads by judging whether the radial blur image is normal is applied differently.

In the present invention, normality can be determined based on the characteristics of a lesion included in a radial blur image, and the specific characteristics (S1, S2, S3) are as follows.

S1 represents the ratio of the size of the lesion area (Area(seg _i )) to the size of the radial blur image (Total Area), S2 represents the size of the lesion area, and S3 represents the number of lesion areas.

According to one embodiment of the present invention, the processor (150) can determine whether the radial blur image is normal or abnormal based on whether the ratio of the size of the lesion region to the size of the radial blur image (S1), the size of the lesion region (S2), or the number of lesion regions (S3) exceeds a threshold value. At this time, the threshold values for S1, S2, and S3 can be set to T1, T2, and T3, respectively, and it is determined as abnormal even if the threshold value for any one characteristic is exceeded.

That is, the processor (150) determines that it is abnormal if S1 > T1 or S2 > T2 or S3 > T3, and determines that it is normal otherwise.

According to one embodiment of the present invention, when the radial blur image is normal, the processor (150) identifies the first candidate optic nerve head with the largest maximum intensity value among the plurality of candidate optic nerve heads as the optic nerve head (S40).

The maximum intensity is the value with the greatest intensity in the intensity distribution confirmed after performing radial blur on each candidate optic nerve head, and there is a maximum intensity value for each candidate optic nerve head.

The processor (150) can compare the maximum intensity values of multiple candidate optic nerve heads and identify the first candidate optic nerve head with the largest maximum intensity value as the optic nerve head.

In addition, by examining the radial blur images, it can be seen that the candidate optic discs are related to their intensity and radius, and the candidate optic disc with the smallest radius among the candidate optic discs has the largest maximum intensity value. Therefore, the candidate optic disc with the smallest radius can be identified as the optic disc.

According to one embodiment of the present invention, when a radial blur image is abnormal, the processor (150) identifies a second candidate optic nerve head having the widest intensity distribution width when the intensity maximum value is 80% among a plurality of candidate optic nerve heads as the optic nerve head (S50).

In the present invention, the geometric characteristics of the optic disc are considered to accurately select the optic disc even when the radial blur image is abnormal. The optic disc is formed by a cup and a rim surrounding the cup, and the edge point of the cup has an elevation of 1/5×ΔZ (the vertical distance between the center of the optic disc and the edge point) above the rim. The center of the optic disc is located at the geometric center of the area with the highest intensity value in the image.

Using geometric model observations, each candidate optic disc is evaluated based on the width of the maximum 80% metric (FW80M), and the second candidate optic disc with the widest width is identified as the optic disc. A specific example is described with reference to Fig. 5.

According to one embodiment of the present invention, the position of the optic nerve head can be effectively identified based on color conversion and blurring.

According to one embodiment of the present invention, it is applicable not only to cases where the fundus image is normal but also to cases where there is an abnormality such as a lesion, so that the position of the optic nerve head can be precisely identified.

FIG. 4 is a drawing illustrating a motion blur performed according to one embodiment of the present invention.

FIG. 4 specifically illustrates performing motion blur as described with reference to S10 of FIG. 3.

According to one embodiment of the present invention, the processor (150) can obtain a plurality of blurred images by performing motion blur on the fundus image at different angles, and obtain a directional blurred image by combining the plurality of blurs.

Referring to FIG. 4, the processor (150) can perform motion blur on the fundus image (410) according to different angle information (420) to obtain multiple blur images (430). At this time, the fundus images (410) are all the same, and the angle information (420) has different angles, so that multiple blur images (430) motion-blurred for each angle are obtained. Then, a directional blur image (440) is obtained by combining the multiple blur images (430).

This process can be expressed mathematically as follows:

)을 컨벌루션한다. In the present invention, motion blur is induced on the input L _* component for better optic disc detection. Uniform motion blur is induced by a motion blur direction kernel (

) is convolved.

는 컨벌루션 연산자이고 θ는 블러링 커널의 방향이다. 회전된 커널을 적용하여 N개의 이미지 시퀀스를 얻을 수 있다. Here

is the convolution operator and θ is the direction of the blurring kernel. By applying the rotated kernel, we can obtain a sequence of N images.

As in mathematical expression 10, the accumulated directional blur image is obtained by combining all the first blur images. The directional blur image provides consistent intensity with noise suppression, and can clearly show the higher intensity area for the optic nerve head candidate.

FIG. 5 is a drawing illustrating an example of identifying the position of the optic nerve head according to one embodiment of the present invention.

With reference to S50 of Fig. 3, a method for identifying an optic nerve head among multiple candidate optic nerve heads displayed in an image illustrated in Fig. 5 (a) is specifically described. At this time, it is assumed that the image of Fig. 5 (a) is judged to be abnormal, and it is informed in advance that the candidate optic nerve head with the highest maximum intensity value is No. 22, and the actual optic nerve head is No. 16.

Fig. 5(a) shows 26 candidate optic discs identified using the extended maximum transform technique on the directional blur image, as described with reference to S20 of Fig. 3.

Figure 5 (b) shows the intensity distribution of candidate optic nerve head number 22, (c) shows the intensity distribution of candidate optic nerve head number 16, and (d) shows the width of the intensity distribution (FW80M value) when it is 80% of the maximum intensity value of each of candidate optic nerve head numbers 1 to 26.

If the candidate optic nerve head with the highest maximum intensity value among the candidate optic nerve heads shown in (a) of Fig. 5 is identified as the optic nerve head, an error occurs because it is determined to be candidate optic nerve head number 22. To prevent such an error, the width of the intensity distribution when the maximum intensity value is 80% among multiple candidate optic nerve heads is considered by reflecting the morphological characteristics of the optic nerve head.

Looking at Fig. 5 (b) and Fig. 5 (c), the maximum intensity value in the intensity distribution is larger in (b), but the width of the intensity distribution when it is 80% of the maximum intensity value is wider in (c).

Therefore, referring to (d) of Fig. 5, the optic nerve head candidate number 16, which has the widest intensity distribution width at 80% of the maximum intensity value, can be identified as the optic nerve head, and the optic nerve head can be accurately identified.

FIG. 6 is a diagram illustrating an overall process for identifying the position of the optic nerve head according to one embodiment of the present invention.

Figure 6 is a schematic diagram illustrating the contents explained with reference to Figures 3 to 5 above, and overlapping contents are borrowed from the contents described above.

When a fundus image (input image) is input, the processor (150) converts the color space of the fundus image into the CIELAB color space (color conversion). Then, a directional blur image is obtained and candidate optic disc candidates are identified through an extended maximum transformation.

Radial blur is performed based on the centers of candidate optic discs to obtain a radial blur image. Then, depending on whether the radial blur image is normal or not, in the case of normality (Abnormality No), the candidate optic disc with the largest maximum intensity value is identified as the optic disc. In the case of abnormality (Abnormality Yes), the candidate optic disc with the widest intensity distribution when it is 80% of the maximum intensity value is identified as the optic disc.

100: Electronic devices
110: Input section
120: Communications Department
130: Display section
140: Storage
150: Processor

Claims (12)

In an electronic device for identifying the position of the optic disc in a fundus image,
Motion blur is performed on the input fundus image at different angles to obtain a directional blur image.
Using the extended maximum transform technique from the above directional blur image, multiple candidate optic discs are identified,
A radial blur is performed based on the center of each candidate optic nerve head on the above directional blur image to obtain a radial blur image.
If the above radial blur image is normal, the first candidate optic nerve head with the largest maximum intensity value among the multiple candidate optic nerve heads is identified as the optic nerve head.
An electronic device including a processor that identifies a second candidate optic nerve head having the widest intensity distribution when the intensity maximum value is 80% among the plurality of candidate optic nerve heads as the optic nerve head when the radial blur image is abnormal. In the first paragraph,
The above processor,
An electronic device that converts the color space of the above fundus image to the CIELAB color space before performing motion blur. In the first paragraph,
The above processor,
The above fundus image is subjected to motion blur at different angles to obtain multiple first blur images,
An electronic device that obtains the directional blur image by combining the plurality of first blur images. In the first paragraph,
The above processor,
A plurality of second blur images are obtained by performing radial blur based on the center of each candidate optic nerve head for the above directional blur image,
An electronic device that obtains the radial blur image by combining the plurality of second blur images. In paragraph 4,
The above processor,
An electronic device for identifying an intensity maximum from the intensity distribution of each candidate optic nerve head obtained through the above radial blur image. In the first paragraph,
The above processor,
An electronic device that determines whether the radial blur image is normal or abnormal based on whether the ratio of the size of the lesion area to the size of the radial blur image, the size of the lesion area, or the number of lesion areas exceeds a threshold value. A method for identifying the position of the optic disc in a fundus image performed by an electronic device,
A step of obtaining a directional blur image by performing motion blur at different angles on an input fundus image;
A step of identifying multiple candidate optic nerve heads using an extended maximum transform technique from the above directional blur image;
A step of performing radial blur on the above directional blur image based on the center of each candidate optic nerve head to obtain a radial blur image;
If the above radial blur image is normal, a step of identifying the first candidate optic nerve head having the largest maximum intensity value among the plurality of candidate optic nerve heads as the optic nerve head;
A method comprising the step of identifying a second candidate optic nerve head having the widest intensity distribution width when the intensity maximum value is 80% among the plurality of candidate optic nerve heads as the optic nerve head when the above radial blur image is abnormal. In Article 7,
The step of obtaining the above directional blur image is:
A method comprising the step of converting the color space of the fundus image into the CIELAB color space before performing motion blur. In Article 7,
The step of obtaining the above directional blur image is:
A step of obtaining a plurality of first blur images by performing motion blur on the above fundus image at different angles;
A method comprising the step of obtaining the directional blur image by combining the plurality of first blur images. In Article 7,
The step of obtaining the above radial blur image is:
A step of performing radial blurring based on the center of each candidate optic nerve head on the above directional blur image to obtain a plurality of second blur images;
A method comprising the step of obtaining the radial blur image by combining the plurality of second blur images. In Article 10,
A method further comprising the step of identifying an intensity maximum from the intensity distribution of each candidate optic nerve head obtained through the above radial blur image. In Article 7,
A method further comprising a step of determining whether the radial blur image is normal or abnormal based on whether the ratio of the size of the lesion area to the size of the radial blur image, the size of the lesion area, or the number of lesion areas exceeds a threshold value.