KR101761510B1 - Apparatus and method for generating fundus image filters for vascular visualization of fundus image - Google Patents

Abstract

An apparatus for generating an eye fundus image filter for visualizing a fundus image and a method thereof are disclosed.
A method of setting an eye fundus image filter comprises: extracting effective pixels of which fundus is displayed in each of a plurality of sample fundus images; Calculating an average of Red, Green, and Blue values of each of the effective pixels; Determining a Red weight, a Green weight, and a Blue weight based on the average of the Red, Green, and Blue values and the average of the absorbance of hemoglobin and the brightness of the plurality of color fundus images; And setting a filter for filtering the color fundus image to apply the Red weight, Green weight, and Blue weight to the color fundus image.

Description

FIELD OF THE INVENTION The present invention relates to an apparatus and method for generating an eye fundus image filter for visualizing a blood vessel of an eye fundus image,

The present invention relates to an apparatus and method for generating fundus image filters for processing fundus images.

Fundus images are one of the most commonly used ophthalmic images for diagnostic or recording purposes in ophthalmology. The fundus image is relatively similar to the fundus of the subject observed at the time of medical examination and is intuitive and thus used for the examination of eye diseases.

Although many diseases in which vascular changes occur in vitreoretinal diseases, funduscopic images do not show blood vessels specifically, so red-free filters and fluorescent funduscopic imaging have been developed to clearly display blood vessels.

The red-free filter, which is a hardware filter, removes the band corresponding to red in the fundus image to display blood vessels with hemoglobin darker than the other regions. Fluorescein is injected into the vein, It is a method of optically photographing a substance circulating through the retinal circulation system to display blood vessels.

However, there is a limitation that a red-free filter requires a camera including a special filter, and fluorescence fundus photography can be limited to a user who can use it due to the specific reaction of the fluorescent material used. In addition, red-free filters and fluorescence fundus photography do not use funduscopic images taken by a general camera, and thus additional tests are required for users who have taken funduscopic images.

Accordingly, there is a demand for a method of displaying blood vessels using funduscopic images taken by a general camera.

The present invention relates to an apparatus and a method for setting or generating an funduscopic image filter for applying Red weight, Green weight, and Blue weight based on the absorbance of retinal artery / vein to color fundus images. Can be provided.

According to another aspect of the present invention, there is provided a fundus image filter generating method comprising: extracting an effective pixel having a fundus of each of a plurality of sample fundus images; Calculating an average of Red, Green, and Blue values of each of the effective pixels; Determining a Red weight, a Green weight, and a Blue weight based on the average of the Red, Green, and Blue values and the average of the absorbance of hemoglobin and the brightness of the plurality of color fundus images; And setting a filter for filtering the color fundus image to apply the Red weight, Green weight, and Blue weight to the color fundus image.

The step of determining weights of the funduscopic image filter generating method according to an exemplary embodiment of the present invention may include determining a weight of a plurality of samples by applying a Red weight, a Green weight, and a Blue weight to the averages of the Red, Green, The Red weight, the Green weight, and the Blue weight can be determined in consideration of the average brightness of the fundus images.

The step of determining the weights of the fundus image filter generating method according to an embodiment of the present invention may determine the ratio between the Blue weight and the Green weight according to the ratio between the average absorbance of Blue of hemoglobin and the average absorbance of Green of hemoglobin have.

The step of determining a weight of the funduscopic image filter generating method according to an embodiment of the present invention includes determining a weight of Red to be applied to a Red value of each effective pixel to reduce a Red value of each of the valid pixels to a predetermined ratio .

The step of determining the weights of the fundus image filter generating method according to an exemplary embodiment of the present invention may include calculating a Green value of each of the effective pixels by considering a rate at which the Red value of each of the effective pixels is reduced by the Red weight, The Green weights can be determined.

According to another aspect of the present invention, there is provided a funduscopic image processing method comprising the steps of: confirming an funduscopic image filter in which a Red weight, a Green weight, and a Blue weight are set according to an average of Red, Green, and Blue values of a plurality of sample fundus images; And applying the Red weight, Green weight, and Blue weight to the color fundus image using the funduscopic image filter to transform the visualized blood vessel into a fundus enhanced gray fundus image.

The Red weight, the Green weight, and the Blue weight of the funduscopic image processing method according to an exemplary embodiment of the present invention are obtained by calculating an average of the red, green, and blue values of each of the effective pixels extracted from each of the plurality of sample fundus images and the average of the hemoglobin The absorbance and the average of the brightness of the plurality of color fundus images.

The funduscopic image filter of the funduscopic image processing method according to an embodiment of the present invention may include the Blue weight and the Green weight in a ratio determined according to the ratio between the average absorbance of Blue of hemoglobin and the average absorbance of Green of hemoglobin.

The Red weight of the funduscopic image processing method according to an exemplary embodiment of the present invention may be applied to the Red value of each of the valid pixels to reduce the Red value of each of the valid pixels to a predetermined ratio.

The Green weight of the funduscopic image processing method according to an embodiment of the present invention can increase the Green value of each of the effective pixels in consideration of the rate at which the Red value of each of the effective pixels is reduced by the Red weight .

According to an embodiment of the present invention, an eye fundus image filter is set up to apply Red weight, Green weight, and Blue weight based on the absorbance of the sample fundus images and the retinal artery / vein to the color fundus image, Or creating a digital filter that increases the frequency and visibility of the retinal blood vessels and is easy to apply to fundus images.

FIG. 1 is a block diagram of an eye fundus image filter generating apparatus according to an embodiment of the present invention. Referring to FIG.
2 is an example of a process of the funduscopic image filter according to an embodiment of the present invention.
3 is an example of filtered fundus image according to an embodiment of the present invention.
4 is an example of effective pixels of the funduscopic image filtered according to an embodiment of the present invention.
5 is an example of a process of extracting an edge of a blood vessel from a filtered funduscopic image according to an embodiment of the present invention.
FIG. 6 is a view showing blood vessels visualized in effective pixels of a funduscopic image filtered according to an embodiment of the present invention.
FIG. 7 illustrates blood vessels visualized in a filtered fundus image according to an embodiment of the present invention. Referring to FIG.
8 is a flowchart illustrating an eye fundus image filter generating method according to an embodiment of the present invention.
9 is a flowchart illustrating an eye fundus image processing method according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The funduscopic image filter generating method according to an embodiment of the present invention can be performed by an funduscopic image filter generating apparatus.

FIG. 1 is a block diagram of an eye fundus image filter generating apparatus according to an embodiment of the present invention. Referring to FIG.

The funduscopic image filter generating apparatus 110 can generate or set an funduscopic image filter for visualizing blood vessels of the funduscopic image using a plurality of sample fundus images.

At this time, the funduscopic image filter generating apparatus 110 may extract effective pixels having fundus images from each of the plurality of sample fundus images. Next, the eye fundus image filter generating apparatus 110 may calculate the average of the Red, Green, and Blue values of each of the extracted effective pixels.

Next, the eye fundus image filter production apparatus 110 determines Red weights, Green weights, and Blue weights based on the average of the Red, Green, and Blue values, the absorbance of hemoglobin, and the average of the brightness of the plurality of color fundus images .

At this time, the funduscopic image filter generating apparatus 110 calculates the red, green, and blue values based on the fact that the result of applying the Red weight, the Green weight, and the Blue weight to the average of the Red, Green, and Blue values is an average of the brightness of the plurality of sample fundus images. Weights, Green weights, and Blue weights.

At this time, the eye fundus image filter generating apparatus 110 can determine the ratio between the Blue weight and the Green weight according to the ratio between the average absorbance of Blue of hemoglobin and the average absorbance of Green of hemoglobin. In addition, the funduscopic image filter generating apparatus 110 may apply the Red value of each valid pixel to determine a Red weight so as to reduce the Red value of each of the valid pixels to a predetermined ratio. In addition, the eye fundus image filter generating apparatus 110 may determine the Green weight so as to increase the Green value of each of the effective pixels in consideration of the rate at which the Red value of each effective pixel is reduced by the Red weight.

Finally, the fundus image filter generating apparatus 110 generates an eye fundus image filter configured to apply Red weight, Green weight, and Blue weight to the color fundus image, or an eye fundus image filter that filters the color fundus image, , Green weight, and Blue weight.

The funduscopic image processing apparatus 120 can visualize the blood vessels included in the funduscopic image by applying the funduscopic image filter created or set by the funduscopic image filter generation apparatus 110 to the color fundus image and converting the same into a gray fundus image .

First, the funduscopic image processing apparatus 120 may receive the funduscopic image filter from the funduscopic image filter generating apparatus 110 or may check the funduscopic image filter previously stored in the funduscopic image processing apparatus 120.

to the next. The funduscopic image processing apparatus 120 can convert the fundus image into the enhanced fundus image by applying Red weight, Green weight, and Blue weight to the color fundus image using the fundus image filter.

The funduscopic image filter generation unit 110 sets an funduscopic image filter that applies Red weight, Green weight, and Blue weight based on the absorbance of the sample fundus images and the retinal artery / vein to the color fundus image, Or creating a digital filter that increases the frequency and visibility of the retinal blood vessels and is easy to apply to fundus images.

2 is an example of a process of the funduscopic image filter according to an embodiment of the present invention.

The brightness L of the image obtained by gray-converting the sample fundus images can be defined as Equation (1).

Pix [i, j] [0] is the Red value of the sample fundus image, and pix [i, j] [1] is the Green value of the sample fundus image. And Pix [i, j] may be a gray-converted image of the sample fundus images.

Also, when the image obtained by gray-converting the sample fundus images is an 8-bit image, the average brightness of the image obtained by gray-converting the sample fundus images may have a value of 0 to 255. [ Accordingly, the funduscopic image filter generating apparatus 110 can set 128, which is an intermediate value between the minimum value 0 and the maximum value 255, to the average brightness of the gray-converted image of the sample fundus images. At this time, equation (2) can be derived.

At this time, Red_avg, Green_avg, and Blue_avg may be the average values of Red, Green, and Blue of the pixels included in the sample fundus images.

The absorbance of Red is significantly lower than that of Blue and Green in the absorbance curve of hemoglobin. On the other hand, among the colors included in the sample fundus image, Red component may be the most. Therefore, the average value of Red values among the Blue average, Green average, and Red average of the pixels included in the sample fundus image may be greatest.

Accordingly, the eye fundus image filter generating apparatus 110 can increase the contrast between Blue and Green by subtracting the Red value 230 by setting the Red weight to a negative value. That is, the eye fundus image filter generating apparatus 110 sets an eye fundus image filter for reducing the Red value 230 from a value obtained by adding the Green value 210 and the Blue value 220 of the color fundus image as shown in FIG. , Or the like, the visibility of blood vessels containing hemoglobin can be increased.

Specifically, the eye fundus image filter generating apparatus 110 can extract the effective pixels from the fundus image. The fundus image may include an effective image with fundus and an external mask without fundus. Therefore, the eye fundus image filter generating apparatus 110 can confirm the number of effective pixels included in the fundus image using the following code.

i, j] [0] + pix [i, j] [1] + pix [i,

                pix_count = pix_count + 1 # number of pixels in mask area count

    Eff_pix = (im.size [0] * im.size [1]) - pix_count # (unknown horizontal pixels x vertical pixels)

The funduscopic image filter generating apparatus 110 may calculate an average value of Red, Green, and Blue of the pixels included in the sample fundus images using Equation (3).

At this time, Red_pix, Green_pix, and Blue_pix may be the sum of values of red, green, and blue in the entire fundus image. For example, red_pix may be a sum of red values of all pixels of the fundus image. Also, eff_pix may be the number of effective pixels extracted from the fundus image by excluding the mast. And float () can be a function to convert Red_pix, Green_pix, Blue_pix to floating point.

For example, the average value of Red of each of the pixels obtained from 400 images is 137.9616, the average value of Green is 89.2767, and the average value of Blue is 9.6764.

Therefore, if Red, Green, and Blue average values are 137.9616, 89.2767, and 9.6764, respectively, Red_avg, Green_avg, and Blue_avg can be converted into Equation (4).

In addition, the funduscopic image filter generating apparatus 110 may convert the red value, the green value, and the blue value into the wavelength (nm), and determine the absorbance of oxygenated hemoglobin and hemoglobin in the converted wavelength band. For example, the absorbances of HbO2 at red (653 nm), green (510 nm), and blue (440 nm) are 340.00, 19040.00 and 68000.00 (cm-1 / M), respectively, and the absorbances of Hb are 3244.00, 23520.00, and 58000.00 1 / M). The mean absorbance of arterial, vein and arteriovenous arteries was calculated as Red (653 nm), Green (653 nm), and Green (R), respectively, when the retinal artery oxygen saturation was 92.2 ± 3.7% and the retinal vein oxygen saturation was 55.6 ± 6.3% (Red: 566.51, Green: 35900.48, 107936.00), (Red: 1629.38, Green: 21029.12, Blue: 63560.00), (Red: 1097.94, Green: 28464.80, Blue: 85748.00) cm < -1 > / M).

Therefore, the ratio of the Blue weight to the Green weight can be expressed by Equation (5).

At this time, the eye fundus image filter generating apparatus 110 can determine the ratio of the blue weight to the green weight according to the ratio between the average absorbance of Blue of hemoglobin and the average absorbance of Green of hemoglobin. At this time, since the ratio of the blue weight is larger than the Green weight, the funduscopic image filter generating unit 110 can convert the Blue weight to a multiple of the Green weight. For example, the funduscopic image filter generating apparatus 110 may derive Equation 6 by transforming the Blue weight value into a multiple of the Green weight value in Equation (4).

In addition, when 10% of the Red value 230 is subtracted, the funduscopic image filter generating apparatus 110 reduces the Red value 230 by correcting the Red weight and the Green weight as shown in Equation (7) The Green value 210 can be increased according to the reduction of the Red value 230. [ At this time, since the Blue value 220 has a constant ratio with the Green value 210, the Blue value 220 can also increase with the increase of the Green value 210.

The funduscopic image filter generating apparatus 110 may determine Red weight, Green weight, and Blue weight as -0.0572, 0.7335, and 2.2079, respectively, using Equations (6) and (7).

3 is an example of filtered fundus image according to an embodiment of the present invention.

The sample fundus image 310 may include an effective image 311 including a fundus and an external mask 312 not including fundus, as shown in FIG. Therefore, the eye fundus image filter generating apparatus 110 can extract the effective pixels of the effective image 311 from the sample fundus image 310.

3 is a gray fundus image obtained by filtering the same color fundus image as the sample fundus image 310 with a conventional green filter and the image 330 of FIG. And a gray fundus image obtained by filtering the same color fundus image as the sample fundus image 310 with the fundus image filter generated by the filtering unit 310. [

The image 340 of FIG. 3 may be a hemoglobin image for use as a reference. Since the fluorescence fundus image is generated by injecting pigment into blood vessels and then capturing the blood vessels, it is difficult to increase the visibility of blood vessels to the degree of fluorescence fundus images only by image processing. However, as shown in FIG. 3, Can be displayed more clearly so that the blood vessels included in the image 340 are close to the image 340.

4 is an example of effective pixels of the funduscopic image filtered according to an embodiment of the present invention.

The image 410 of FIG. 4 may be an image showing only valid pixels of the same color fundus image as the sample fundus image 310. 4 is a gray fundus image obtained by filtering the image 410 with a conventional green filter and the image 430 of FIG. 4 is an eye fundus image filter generated by the fundus image filter generating apparatus 110 And may be a gray fundus image obtained by filtering the image 410. In addition, the image 440 of FIG. 4 may be a hemoglossal fundus image for use as a reference.

4, a blood vessel included in the image 430 obtained by filtering the image 410 with the funduscopic image filter generated by the funduscopic image filter generating apparatus 110 is displayed on the image 440 As shown in FIG.

5 is an example of a process of extracting an edge of a blood vessel from a filtered funduscopic image according to an embodiment of the present invention.

The funduscopic image processing apparatus 120 may apply edge detection to the gray fundus image 510 to measure the quality of the gray fundus image 510 filtered through the funduscopic image filter. For example, the funduscopic image processing apparatus 120 may apply an edge detection process to the gray fundus image 510 using a FIND_EDGES filter (convolution with 2D laplacian kernel) of a python image module. The funduscopic image processing apparatus 120 may store the image 520 to which the edge detection process is applied to the gray fundus image 510 together with the gray fundus image 510.

In addition, the funduscopic image processing apparatus 120 generates an image 520 using a convolution with a 2D Laplacian kernel, and measures the frequency of the boundary line of the blood vessels included in the image 520 .

FIG. 6 is a view showing blood vessels visualized in effective pixels of a funduscopic image filtered according to an embodiment of the present invention.

The funduscopic image processing apparatus 120 converts the color fundus image 610 into a gray fundus image 620 using an eye fundus image filter applying Red weight, Green weight, and Blue weight to the color fundus image 610, The boundaries of the blood vessels 621 can be made clear as shown in FIG.

FIG. 7 illustrates blood vessels visualized in a filtered fundus image according to an embodiment of the present invention. Referring to FIG.

The funduscopic image processing apparatus 120 converts the color fundus image 710 into a gray fundus image 720 using an eye fundus image filter applying Red weight, Green weight, and Blue weight to the color fundus image 710, The boundaries of the blood vessels 721 can be made clear as shown in FIG.

8 is a flowchart illustrating an eye fundus image filter generating method according to an embodiment of the present invention.

In step 810, the fundus image filter generating apparatus 110 may extract an effective pixel whose fundus is displayed in each of the plurality of sample fundus images.

In step 820, fundus image filter generating apparatus 110 may calculate an average of Red, Green, and Blue values of each of the valid pixels extracted in step 810.

In step 830, the fundus image filter generating apparatus 110 generates a red weight based on the average of the absorbance of hemoglobin, the average of the brightness of the plurality of color fundus images, and the average of the Red, Green, and Blue values calculated in step 820. [ , Green weights, and Blue weights. At this time, the funduscopic image filter generating apparatus 110 calculates the red, green, and blue values based on the fact that the result of applying the Red weight, the Green weight, and the Blue weight to the average of the Red, Green, and Blue values is an average of the brightness of the plurality of sample fundus images. Weights, Green weights, and Blue weights.

In step 840, the fundus image filter generating apparatus 110 generates an eye fundus image filter configured to apply a Red weight, a Green weight, and a Blue weight to the color fundus image, or an eye fundus image filter that filters a color fundus image, Can be set to apply Red weights, Green weights, and Blue weights.

9 is a flowchart illustrating an eye fundus image processing method according to an embodiment of the present invention.

In step 910, the funduscopic image processing apparatus 120 may receive a color fundus image. At this time, the funduscopic image processing apparatus 120 may receive the funduscopic image filter from the funduscopic image filter generating apparatus 110, or may check the funduscopic image filter previously stored in the funduscopic image processing apparatus 120.

In step 910, the funduscopic image processing apparatus 120 can convert the fundus image into the enhanced fundus image by applying Red weight, Green weight, and Blue weight to the color fundus image.

The present invention establishes or creates an eye fundus image filter that applies Red weight, Green weight, and Blue weight based on the sample fundus images and the absorbance of the retinal artery / vein to the color fundus image, And the digital filter which can be easily applied to the funduscopic image can be derived.

The method according to an embodiment may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer readable medium may include program instructions, data files, data structures, and the like, alone or in combination. Program instructions to be recorded on the medium may be those specially designed and constructed for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the equivalents of the claims, as well as the claims.

110: fundus image filter generating device
120: fundus image processing device

Claims (10)

Extracting an effective pixel whose fundus is displayed in each of the plurality of sample fundus images;
Calculating an average of Red, Green, and Blue values of each of the effective pixels;
Determining a Red weight, a Green weight, and a Blue weight based on the average of the Red, Green, and Blue values and the average of the absorbance of hemoglobin and the brightness of the plurality of color fundus images; And
Generating an eye fundus image filter applying the Red weight, Green weight, and Blue weight to the color fundus image;
/ RTI > of the fundus image filter. The method according to claim 1,
The step of determining the weight includes:
Green weight, and Blue weight, considering that the result of applying Red weight, Green weight, and Blue weight to the average of the Red, Green, and Blue values is an average of the brightness of a plurality of sample fundus images, Determining a fundus image filter. The method according to claim 1,
The step of determining the weight includes:
Wherein the ratio between the Blue weight and the Green weight is determined according to a ratio between an average absorbance of Blue of hemoglobin and an average absorbance of Green of hemoglobin. The method according to claim 1,
The step of determining the weight includes:
Wherein the Red weight is determined so as to reduce the Red value of each of the effective pixels to a predetermined Red ratio. 5. The method of claim 4,
The step of determining the weight includes:
Wherein the Green weight is determined so as to increase a Green value of each of the effective pixels in consideration of a rate at which the Red value of each of the effective pixels is reduced by the Red weight. Identifying an eye fundus image filter having a Red weight, a Green weight, and a Blue weight according to an average of Red, Green, and Blue values of a plurality of sample fundus images; And
Applying the Red weight, Green weight, and Blue weight to the color fundus image with the fundus image filter to convert the visualized blood vessel into a fundus enhanced gray fundus image;
/ RTI > The method according to claim 6,
The Red weights, Green weights, and Blue weights,
Green, and blue values of each of the valid pixels extracted from each of the plurality of sample fundus images, the absorbance of hemoglobin, and the average of the brightness of the plurality of color fundus images. The method according to claim 6,
The funduscopic image filter
The Blue weight and the Green weight in a ratio determined by a ratio between an average absorbance of Blue of hemoglobin and an average absorbance of Green of hemoglobin and the Green weight. 8. The method of claim 7,
The Red weight is a weight
Wherein the Red value of each of the valid pixels is applied to the Red value of each of the valid pixels to reduce the Red value of each of the valid pixels to a predetermined ratio. 10. The method of claim 9,
The Green weight value
Wherein the Green value of each of the effective pixels is increased in consideration of a rate at which the Red value of each of the effective pixels is reduced by the Red weight.

Images