JP2017023482A - Picture processing device, picture processing method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a mechanism capable of accurately detecting a visual cell of a subject's eye.SOLUTION: A picture processing device comprises: a picture acquisition unit 110 that acquires a picture obtained by imaging a visual cell of a subject's eye; a division unit 142 that divides the picture acquired by the picture acquisition unit 110 into small areas; a feature amount acquisition unit 143 that performs space frequency conversion on each of the small areas obtained by the division unit 142's division to acquire a feature amount of each small area; and an area extraction unit 144 that, on the basis of the feature amount of each small area acquired by the feature amount acquisition unit 143, extracts a non-resolution area, i.e., an area in which a visual cell has not yet been resolved, in the picture acquired by the picture acquisition unit 110.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an image processing apparatus and an image processing method for processing an image obtained by photographing a photoreceptor cell of an eye to be examined, and a program for causing a computer to execute the image processing method.

  The visual cells in the retina of the eye play an important role in the visual function, which is an important function in the human senses. In recent years, by applying aberration correction technology, a technique for drawing the fundus retina of the eye to be examined with high definition has been constructed, and discovery of new knowledge about visual functions and construction of diagnostic value are expected.

  An adaptive optics scanning laser opthalmoscope (hereinafter referred to as “AO-SLO”), which applies adaptive optics, employs a telescope technique for acquiring a clear star image by compensating for atmospheric fluctuations. This is an ophthalmoscope applied to. With this AO-SLO device technology, it has become possible to resolve each photoreceptor cell of the retina of the eye to be examined. It is said that photoreceptor cells are present at intervals of about 2 μm in the vicinity of the fovea, which is the highest density. Currently, the development for improving the resolution is continued for the purpose of drawing the photoreceptor cells in the fovea.

  In order to connect an image acquired by the AO-SLO device to clinical value, it is necessary to extract an index effective for diagnosis from the image. One technique required for this purpose is to accurately detect the visualized visual cells and evaluate the density in association with the distance from the fovea. It is known that the density of photoreceptor cells changes according to the distance from the fovea, and the density near the fovea is the highest. The size of the density varies even in a normal eye, and it may be difficult to depict individual photoreceptor cells near the fovea even with a high-resolution AO-SLO device.

  As a conventional technique, Non-Patent Document 1 describes a technique of paying attention to a luminance profile of a photoreceptor cell, and detecting a maximum value after low-pass filter processing corresponding to a frequency corresponding to a photoreceptor cell cycle. Further, Non-Patent Document 2 describes a technique for determining a position with the highest density from a change in the density of photoreceptor cells and using it as a fovea.

JP2013-63215A

K. Y. Li and A. Roorda, "Automated identification of cone subunits in adaptive optics retinal images," J. Opt. Soc. Am. A 24 (5), 1358-1363 (2007). K. Y. Li, P. Tiruveedhula and A. Roorda, "Intersubject Variability of Foveal Cone Photoreceptor Density in Relation to Eye Length," IOVS 2010; 51: 6858-67 Curcio CA, Sloan KR, Kalina RE, Hendrickson AE.Human replica topography.J Comp Neurol. 1990; 292: 497-523.

  In the technique described in Non-Patent Document 1, when the density of photoreceptor cells increases in the vicinity of the fovea, the resolution limit of the AO-SLO device is approached, and adjacent photoreceptor cells are resolved to individually identify photoreceptor cells. It becomes difficult. In this case, various statistical indices such as photoreceptor cell density and closest distance, and the foveal position acquired by the technique described in Non-Patent Document 2 are techniques that depend on the photoreceptor detection results, and therefore are accurate values. It will not be.

  Patent Document 1 describes a technique for extracting a low-quality area from an image of photoreceptor cells. However, the technique described in Patent Document 1 cannot perform processing suitable for the situation when the discrimination of photoreceptor cells becomes insufficient due to the high density of photoreceptor cells even if the image quality is high. It was difficult to accurately detect cells.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a mechanism capable of accurately detecting photoreceptor cells of an eye to be examined.

An image processing apparatus according to the present invention includes an image acquisition unit that acquires an image obtained by photographing a photoreceptor cell of an eye to be examined, a dividing unit that divides the image into small regions, and a spatial frequency that is obtained by dividing each small region by the dividing unit. Based on the feature amount acquisition means for converting and acquiring the feature amount of each small region, an unresolved region in which the photoreceptor cell is an unresolved region in the image based on the feature amount of each small region Region extracting means for extracting.
The present invention also includes an image processing method by the above-described image processing apparatus and a program for causing a computer to execute the image processing method.

  According to the present invention, it is possible to accurately detect photoreceptor cells in the eye to be examined.

It is a block diagram which shows an example of schematic structure of the ophthalmologic apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the 1st Embodiment of this invention and shows the some AO-SLO image and WF-SLO image which were produced | generated by the AO-SLO apparatus of FIG. It is a figure which shows the 1st Embodiment of this invention and shows the fixation lamp map for operating the lighting position of the fixation lamp in the AO-SLO apparatus of FIG. It is a flowchart which shows an example of the process sequence of the image processing method by the image processing apparatus which concerns on the 1st Embodiment of this invention. It is a flowchart which shows an example of the detailed process sequence of the feature-value acquisition process in FIG.4 S160. FIG. 5 is a diagram illustrating a method for calculating an integration function from the spatial frequency image Fi of each small region obtained in step S150 in FIG. 4 according to the first embodiment of this invention. It is a figure which shows the 1st Embodiment of this invention and shows an example of the shape of an integration function. It is a figure which shows the 1st Embodiment of this invention and shows an example of the process at the time of extracting an example of a peak ratio image and an unresolved area | region. It is a flowchart which shows an example of the process sequence of the image processing method by the image processing apparatus which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows an example of the process sequence of the image processing method by the image processing apparatus which concerns on the 3rd Embodiment of this invention. It is a flowchart which shows an example of the process sequence of the image processing method by the image processing apparatus which concerns on the 4th Embodiment of this invention.

  Hereinafter, embodiments (embodiments) for carrying out the present invention will be described with reference to the drawings.

(First embodiment)
First, a first embodiment of the present invention will be described.

  In the first embodiment, from an AO-SLO image obtained by photographing a retina (including photoreceptor cells) of an eye to be examined by an AO-SLO apparatus, between adjacent photoreceptor cells in a region where the photoreceptor cells around the fovea of the eye to be examined are dense. A method for extracting a region in which it is difficult to discriminate photoreceptor cells individually because resolution of the above cannot be performed will be described. In the following description, an area where it is difficult to individually discriminate photoreceptor cells (that is, an area where photoreceptor cells are unresolved in an AO-SLO image) is referred to as an “unresolved area”.

  Specifically, in the present embodiment, an AO-SLO image obtained by photographing the retina (including photoreceptor cells) of the eye to be examined is divided into small areas, and spatial frequency conversion processing is performed on each divided small area. Then, in the present embodiment, based on the features of the spatial frequency converted image of each small region obtained by this spatial frequency conversion processing, a feature amount indicating the resolution of each small region is acquired, and these feature amounts are used for the entire image. To un-resolved regions in the AO-SLO image. Thus, by using the features of the spatial frequency converted image reflecting the arrangement of photoreceptor cells in the AO-SLO image, it becomes possible to extract an unresolved region by a method that does not depend on the detection accuracy of photoreceptor cells. .

[Schematic configuration of ophthalmic apparatus]
FIG. 1 is a block diagram showing an example of a schematic configuration of an ophthalmologic apparatus 10 according to the first embodiment of the present invention.
As shown in FIG. 1, the ophthalmologic apparatus 10 according to the present embodiment includes an image processing apparatus 100, an AO-SLO apparatus 200, an information input apparatus 300, a display apparatus 400, and a database 500.

  The AO-SLO device 200 is a device that captures a retina (including photoreceptor cells) of a subject eye and generates a planar image such as an AO-SLO image or a wide-angle WF-SLO image.

  The information input device 300 is a device that inputs various types of information to the image processing apparatus 100.

  The image processing apparatus 100 is an apparatus that processes a planar image such as an AO-SLO image or a WF-SLO image obtained by photographing the retina (including photoreceptor cells) of the eye to be examined by the AO-SLO apparatus 200. As shown in FIG. 1, the image processing apparatus 100 includes an image acquisition unit 110, an information acquisition unit 120, a control unit 130, an image processing unit 140, a storage unit 150, and an output unit 160. .

  The image acquisition unit 110 performs processing for acquiring a planar image such as an AO-SLO image or a WF-SLO image obtained by photographing the retina (including photoreceptor cells) of the eye to be examined by the AO-SLO device 200.

  The information acquisition unit 120 performs processing for acquiring various types of information from the AO-SLO device 200 and the information input device 300. For example, the information acquisition unit 120 acquires information about the eye to be examined from the AO-SLO apparatus 200 or acquires input information from the examiner from the information input apparatus 300.

  The control unit 130 performs overall control of operations in the image processing apparatus 100 and performs various processes in the image processing apparatus 100.

  The image processing unit 140 processes a planar image such as an AO-SLO image or a WF-SLO image acquired by the image acquisition unit 110. As illustrated in FIG. 1, the image processing unit 140 includes a positioning unit 141, a dividing unit 142, a feature amount acquiring unit 143, a region extracting unit 144, and a photoreceptor analyzing unit 145.

The alignment unit 141 performs alignment processing between the AO-SLO image acquired by the image acquisition unit 110 and the WF-SLO image.
The dividing unit 142 performs a process of dividing the AO-SLO image that has been aligned by the positioning unit 141 into small regions.
The feature amount acquisition unit 143 performs a process of performing spatial frequency conversion on each small region divided by the dividing unit 142 and acquiring the feature amount of each small region. Specifically, in the present embodiment, the feature amount acquisition unit 143 performs a process of acquiring a feature amount based on the shape of an integration function obtained by integrating the absolute value of the output in each angle direction for each small region.
The region extraction unit 144 performs a process of extracting an unresolved region in which the photoreceptor cell is an unresolved region in the AO-SLO image based on the feature amount of each small region acquired by the feature amount acquisition unit 143.
The photoreceptor analysis unit 145 performs photoreceptor cell analysis processing on the AO-SLO image.

  The storage unit 150 stores programs and various information necessary for the control unit 130 to perform processing. Further, based on the control of the control unit 130, the storage unit 150 obtains a planar image acquired by the image acquisition unit 110, various types of information acquired by the information acquisition unit 120, and further, results of processing of the image processing unit 140. Various kinds of information are stored.

  The output unit 160 displays the planar image acquired by the image acquisition unit 110, various types of information acquired by the information acquisition unit 120, and various types of information obtained as a result of processing by the image processing unit 140, etc. Are output and displayed on the database 500, and are also output and stored in the database 500.

  The display device 400 performs processing for displaying a planar image, various information, and the like output from the output unit 160.

  The database 500 performs processing for storing the planar image output from the output unit 160, various information, and the like.

[Plane image]
FIG. 2 is a diagram schematically illustrating a plurality of AO-SLO images and WF-SLO images generated by the AO-SLO apparatus 200 of FIG. 1 according to the first embodiment of this invention.
In the AO-SLO device 200, it is possible to photograph different positions of the retina of the subject's eye by changing the lighting position of the fixation lamp and photographing the subject as if the subject's eye is staring at different positions. Here, FIG. 2 shows a plurality of AO-SLO images 220 and a wide-angle WF-SLO image 210 generated by the AO-SLO device 200.

FIG. 3 is a diagram showing a fixation lamp map for operating the fixation lamp lighting position in the AO-SLO device 200 of FIG. 1 according to the first embodiment of the present invention.
First, the AO-SLO device 200 turns on the fixation lamp with the center position selected in the fixation lamp map shown in FIG. Hereinafter, this position is referred to as a reference position. At this time, if the subject's eye staring at the fixation light to be presented is photographed, the vicinity of the macula of the subject's eye can be photographed.

  Here, the WF-SLO image 210 shown in FIG. 2 is a low-resolution image acquired without using the compensation optical system in an optical system different from the case where the AO-SLO image is acquired by the AO-SLO apparatus 200. There is an image for obtaining an entire image of the retina by photographing a wide range of the retina of the eye to be examined. As shown in FIG. 2, by associating with the WF-SLO image 210, the position of the AO-SLO image 220 having a narrow angle of view in the entire retina is indicated. Hereinafter, in the present embodiment, it is assumed that the WF-SLO image 210 has an image size of 8 mm × 6 mm and a pixel size of 533 pixels × 400 pixels. In addition, the AO-SLO image 220 has three types of shooting area sizes of 1.7 mm × 1.7 mm, 0.82 mm × 0.82 mm, and 0.34 mm × 0.34 mm. There are three types of resolution that are 400 pixels × 400 pixels in common. Here, an AO-SLO image with a shooting area size of 1.7 mm × 1.7 mm is an L image 220 -L, an AO-SLO image with a size of 0.82 mm × 0.82 mm is an M image 220 -M, 0 An AO-SLO image having a size of 34 mm × 0.34 mm is referred to as an S image 220-S. In the photoreceptor cell analysis described below, it is assumed that the S image 220-S is mainly an analysis target. Further, the AO-SLO image 220 and the WF-SLO image 210 are referred to as planar images.

[Shooting protocol]
The imaging protocol of the eye to be examined varies depending on the disease or the like of the eye to be focused on. However, as a standard protocol, first, a WF-SLO image 210 centered on the macula is photographed, and then a plurality of positions of the retina have different resolutions. Photographing is performed while combining the AO-SLO images 220. In this embodiment, it is possible to change the shooting time and frame rate of the AO-SLO image 220. Here, the frame rate is 32 frames per second, the shooting time is 1 second, and the image is composed of 32 images. Shall.

[Processing Procedure of Image Processing Apparatus 100]
FIG. 4 is a flowchart showing an example of the processing procedure of the image processing method by the image processing apparatus 100 according to the first embodiment of the present invention.

<Step S110>
First, in step S <b> 110, the information acquisition unit 120 acquires information about the eye to be examined as patient information from the information input device 300. Then, the information acquisition unit 120 stores the acquired information about the eye to be examined in the storage unit 150 via the control unit 130. Here, the information of the subject's eye includes subject information such as the subject's ID and date of birth, information of measurement data such as the axial length of the subject's eye, and images taken in the past Information is included.

<Step S120>
Subsequently, in step S <b> 120, the image acquisition unit 110 acquires a planar image (WF-SLO image 210 and AO-SLO image 220) in the retina of the eye to be inspected generated by photographing with the AO-SLO device 200. Here, it is assumed that the planar image includes photoreceptor cells in the retina of the eye to be examined. Then, the image acquisition unit 110 stores the acquired planar image in the storage unit 150 via the control unit 130.

<Step S130>
Subsequently, in step S130, the alignment unit 141 performs alignment of the AO-SLO image 220 and the WF-SLO image 210 acquired in step S120. Then, for example, the alignment unit 141 approximates the AO-SLO image 220 based on the relationship between the approximate position of the fovea of the subject eye specified on the WF-SLO image 210 and the aligned AO-SLO image 220. Find the relationship with the fovea position. For example, the alignment unit 141 selects and acquires an AO-SLO image including the fovea in the image based on the obtained relationship.

  Here, there are a plurality of methods for specifying the approximate fovea position on the WF-SLO image 210. For example, since the WF-SLO image 210 is an image obtained by fixing the reference position, there is a method in which the center of the WF-SLO image 210 is an approximate fovea. In addition, there is a method in which the WF-SLO image 210 is analyzed, and the approximate position of the fovea is defined and specified from the running of the blood vessel near the macula of the eye to be examined. Then, the alignment unit 141 controls the result of the association between the AO-SLO image 220 and the WF-SLO image 210 and the AO-SLO image 220 including the approximate fovea position selected based on the result. The data is stored in the storage unit 150 via the unit 130.

<Step S140>
Subsequently, in step S140, the dividing unit 142 performs a process of dividing the AO-SLO image including the fovea acquired in step S130 into small regions. Then, the dividing unit 142 stores information related to the size and position of the divided small areas in the storage unit 150 via the control unit 130. Here, since each small region is a unit for performing spatial frequency conversion, it is necessary to optimize according to the number of photoreceptor cells contained in each small region and the distance from the fovea. Specifically, when each small region is small and the number of photoreceptor cells contained in the small region is small, a correct spatial frequency value cannot be obtained, and conversely, each small region is large and the photoreceptor cell is within the small region. When the density changes, the value of the spatial frequency corresponding to the density of the photoreceptor cells cannot be obtained from the result of the spatial frequency conversion.

  In the example shown below, description is made assuming that the size of the small area is 100 pixels × 100 pixels, but in the present embodiment, the size of the small area is not limited to this value. Further, in the example shown below, the analysis areas are set in a tile shape by shifting the divided small areas up, down, left, and right by 50 pixels. In the present embodiment, the amount of 50 pixels used here can be freely changed. However, when a small amount of shift is set and a large number of small regions are set, analysis with high accuracy is possible, but the calculation load increases. In the example shown below, the small area of 100 pixels × 100 pixels is shifted by 50 pixels in the vertical and horizontal directions for the entire image, and the total is analyzed as 7 × 7 = 49 small areas.

<Step S150>
Subsequently, in step S150, the feature amount acquisition unit 143 performs a spatial frequency conversion process on each small region divided in step S140. Specifically, the feature amount acquisition unit 143 calculates an average luminance value for a small area Ri of 100 pixels × 100 pixels, subtracts it from each pixel value of the small area Ri, and further performs edge processing using a window function. Is arranged in the center of an image Li of 128 pixels × 128 pixels. Although various types of window functions can be considered here, a Hann window function of size 4 is used here. In addition, the feature amount acquisition unit 143 performs padding with the average value calculated above for a region where the pixel value of the small region Ri does not exist in the peripheral portion of the image Li. Then, the feature amount acquisition unit 143 converts the image Li having a size of 128 × 128 pixels into a spatial frequency image Fi by fast Fourier transform (hereinafter referred to as “FFT”). Here, the spatial frequency image Fi is an image in which only the absolute value of the complex number after the spatial frequency conversion process is assigned to each pixel, and the four quadrants of the image are moved so that the pixel corresponding to the low frequency is at the center of the image. doing. Similar processing is applied to all 49 small regions. Then, the feature amount acquisition unit 143 stores the spatial frequency image Fi of each small region Ri thus obtained in the storage unit 150 via the control unit 130.

<Step S160>
Subsequently, in step S160, the feature amount acquisition unit 143 acquires a feature amount for determining whether the image is resolved or unresolved using the spatial frequency image Fi of each small region obtained in step S150. Detailed processing in step S160 will be described with reference to FIG.

  FIG. 5 is a flowchart illustrating an example of a detailed processing procedure of the feature amount acquisition processing in step S160 of FIG.

<< Step S161 >>
First, in step S161, the feature amount acquisition unit 143 calculates an integration function based on the spatial frequency image Fi of each small region obtained in step S150. Here, it is known that a ring-shaped structure corresponding to the periodicity in the arrangement of the photoreceptor cells of the eye to be examined is seen in the spatial frequency image Fi, and the integration function has the magnitude of the frequency corresponding to this periodicity. Calculated for the purpose of obtaining

FIG. 6 shows the first embodiment of the present invention, and is a diagram for explaining a method for calculating the integration function from the spatial frequency image Fi of each small region obtained in step S150 of FIG.
Specifically, as shown in FIG. 6A, a polar coordinate display (r, θ) with the center point (64, 64) of the spatial frequency image Fi as the origin is considered, and the value of each pixel of the spatial frequency image Fi is considered. An integration function I (r), which is a function of integrating in the θ direction, is calculated. However, r = 0, 1, 2,..., N / 2. FIG. 6B shows an integration function I (r) calculated based on the spatial frequency image Fi shown in FIG. The value of the integration function I (r) corresponds to a value obtained by integrating components in all directions (each angle direction) having a frequency magnitude r, and indicates the magnitude of periodicity seen in the image.

<< Step S162 >>
Subsequently, in step S162, the feature amount acquisition unit 143 acquires the peak position r_normal in accordance with the distance from the fovea of the eye to be examined in each small region. Further, the feature amount acquisition unit 143 acquires the peak position in the shape of the integration function I (r) from the integration function I (r) calculated in step S161. Here, the peak position in the shape of the integration function I (r) is related to the density of photoreceptor cells in order to reflect the periodicity of the arrangement of photoreceptor cells. Specifically, when the density of photoreceptor cells is D [count / mm ² ], assuming that photoreceptor cells are arranged in an ideal hexagonal honeycomb, the photoreceptor cell density D is the photoreceptor cell density. The distance L between them is the relationship shown in the following equation (1).

  Furthermore, when the distance between photoreceptor cells is L, the magnitude r of the frequency shown on the spatial frequency image Fi has the relationship shown in the following equation (2).

  Here, the image size (image_size) shown in the equation (2) represents the size of an image to be subjected to FFT in units of μm. At this time, considering the case where the pixel resolution of the small region Ri is 0.85 μm / pixel and the pixel size of the image Li generated by padding around the small region Ri is 128 × 128, the image size is 0.85. × 128 is 108.8 μm.

It is known that the density D of photoreceptor cells changes depending on the distance from the fovea of the retina of the eye to be examined. For example, the density value of photoreceptor cells obtained anatomically in Non-Patent Document 3 and The relationship with the distance from the fovea is shown. Therefore, the feature amount acquisition unit 143 obtains the distance from the fovea of each small region based on the approximate position of the fovea acquired in step S130, and calculates the density of the corresponding photoreceptor cell as described in Non-Patent Document 3. Estimate using. Then, the feature amount acquisition unit 143 acquires the peak position r_normal in accordance with the distance from the fovea of the eye to be examined in each small region, using Expressions (1) and (2). For example, at a position where the distance from the fovea on the nasal side is about 200 μm, the density D of photoreceptor cells is 74000 [count / mm ² ], and the distance L between photoreceptor cells in this case uses equation (1). And about 3.4 μm. At that time, the magnitude r of the frequency on the spatial frequency image Fi is about 31.8 from the equation (2).

Next, a method for obtaining the peak position in each small region will be described.
FIG. 7 is a diagram illustrating an example of the shape of the integration function according to the first embodiment of this invention. In FIG. 7, the horizontal axis represents the frequency magnitude r as in FIG. 6B, and the vertical axis represents the integration function I (r) as in FIG. 6B.

  Specifically, type 1 in FIG. 7 is a type having one peak in the high frequency region, type 2 in FIG. 7 is a type having two peaks, and type 3 in FIG. 7 is one type in the low frequency region. Is a type with When the photoreceptor cells are clearly resolved one by one, a peak appears at a position corresponding to r_normal which reflects the density D of the photoreceptor cells, but the resolution becomes weak and separation between photoreceptor cells present in the vicinity When the state becomes difficult, a peak appears in a low frequency region corresponding to a state where a plurality of photoreceptor cells are fused. Therefore, here, the high frequency region is a region in the vicinity of r_normalative and a region on the high frequency side thereof, and the low frequency region is a region on the low frequency side of about 0.75 of r_normal.

Considering the relationship as described above, in step S162, the feature amount acquisition unit 143 acquires the peak position ri from the integration function of each small region. At this time, when there are two peaks as in type 2 in FIG. 7, the positions of the two peaks are acquired, the peak position on the high frequency side is ri_high, and the peak position on the low frequency side is ri_low. In addition, when only one peak exists as in type 1 and type 3 in FIG. 7, the obtained peak positions r_peak and r_normal are compared, and the frequency is higher than 0.75 times r_normal. If there is, the peak position is ri_high, and if the frequency is low, the peak position is ri_low. For example, in type 1, ri_high = r_peak and ri_low = 0, and in type 3, ri_high = 0 and ri_low = r_peak.
In this way, when there is one peak of the integration function acquired in each small region, the peak is based on the magnitude relationship with r_normalative which is the peak position obtained based on the distance from the fovea of each small region. Are classified into two types. Although an example using the method of acquiring the maximum value of the integration function has been described here as a method of acquiring the peak position ri, the present embodiment is not limited to this method.

<< Step S163 >>
Subsequently, in step S163, the feature amount acquisition unit 143 includes a plurality of types of peaks in the image acquired in step S130 based on the peak position ri in the integration function of each small region acquired in step S162. Judge whether to do. Specifically, in each region of the image acquired in step S130, when both ri_high and ri_low exist, it is determined that a plurality of types of peaks are mixed (S163 / Yes), The process proceeds to step S164. Here, the case where both ri_high and ri_low exist at the peak position ri may be a case where there is a small region having two peaks, or a small region having only ri_high and a small region having only ri_low. The area may exist in the same image. Further, when only one of ri_high and ri_low exists at the peak position ri when viewed in the entire image acquired in step S130, it is determined that a plurality of types of peaks are not mixed (S163 / No), the process proceeds to step S167.

<< Step S164 >>
In step S164, the feature amount acquisition unit 143 calculates a first threshold value that is an intermediate value for an area having two peak positions of the integration function acquired in step S162.
Specifically, only an area having two peak positions is selected, and an average value r_high_ave of the peak position on the high frequency side and an average value r_low_ave of the peak position on the low frequency side are obtained, and an intermediate value is obtained. The first threshold (r_threshold) is assumed.

<< Step S165 >>
Subsequently, in step S165, the feature amount acquisition unit 143 calculates the peak ratio of each small region using the first threshold value calculated in step S164. Specifically, when the small region has two peaks, the feature amount acquisition unit 143 divides the value of the integration function of the high frequency peak position by the value of the integration function of the low frequency peak position, Calculated as the peak ratio of the small region. In addition, when the small region has only one peak and the position of the peak is a high frequency region, the feature amount acquisition unit 143 determines the value of the integration function of the detected peak position in r_low_ave obtained in step S164. The value divided by the value of the integration function is calculated as the peak ratio of the small area. Further, when the small region has only one peak and the peak position is the low frequency region, the feature amount acquisition unit 143 uses the value of the integration function of the peak position detected as the integration function value of r_high_ave. The divided value is calculated as the peak ratio of the small region.
In summary, the peak ratio (peak_ratio) is expressed by the following equation (3).

<< Step S166 >>
Subsequently, in step S166, the feature amount acquisition unit 143 generates a peak ratio image based on the peak ratio of each small region calculated in step S165. Specifically, the feature amount acquisition unit 143 arranges the peak ratio obtained from each small region at the position of the center pixel of the image of each small region, and the value of the pixel value of the entire image is set. Determine from. Here, a pixel value of 400 × 400 is estimated from a given peak ratio value of 7 × 7 = 49 points. A plurality of such methods are known. For example, a method of simply performing linear interpolation between points, or a higher-order interpolation method such as polynomial interpolation or spline interpolation when simple linear interpolation is too coarse can be considered.

Here, an example of performing interpolation by Gaussian will be described as a method for increasing the influence of the peak ratio in the neighboring region, but also taking into account the influence of the peak ratios other than the nearest neighbor.
A two-dimensional Gaussian having a variance σ is defined in an area where the kernel size is about four times the variance σ. Here, the Gaussian kernel is assumed to be standardized, and the total integrated value in the area where the kernel is defined is 1.0. Then, the defined Gaussian is set so that the center of the Gaussian comes on the center pixel of each small area. For the entire image, 49 Gaussians are set, but for each pixel in the image, for each affected Gaussian (a Gaussian whose center is closer than half the kernel size), the respective peak ratio Get value and weight. Here, the weight is a standardized Gaussian kernel value. Then, a smoothing filter is applied to smooth the whole. Thus, for each pixel, the weighted average value of the peripheral peak ratio is calculated as a peak ratio image. FIG. 8A shows an example of the peak ratio image.

<< Step S167 >>
Next, a process when it is determined in step S163 that plural types of peaks are not mixed in the image (S163 / No) will be described. That is, when there is one kind of small area peak in the image, the process proceeds to step S167.
In step S167, the feature amount acquisition unit 143 determines whether the AO-SLO image acquired in step S130 is a resolution when all the small region peaks have only one of ri_high and ri_low. It is determined whether it is. Specifically, the feature amount acquisition unit 143 determines that the AO-SLO image acquired in step S130 has been resolved in all regions when the peak positions of all the small regions are ri_high. . When the peak positions of all small areas are ri_low, it is determined that the AO-SLO image acquired in step S130 is unresolved in all areas. Here, in consideration of the processing so far, the determination processing in step S167 is performed when the peak of the integration function obtained in each small region in the image is one (S163 / No). This corresponds to the process of determining whether the AO-SLO image acquired in step S130 is resolved or unresolved based on the type.

<< Step S168 >>
When the process of step S166 is completed, or when the process of step S167 is completed, the process proceeds to step S168.
In step S168, the feature amount acquisition unit 143 stores the information obtained so far in the storage unit 150 through the control unit 130 as the feature amount of each small region divided in step S140. Specifically, when the process proceeds to step S168 through steps S161 to S166, the feature amount acquisition unit 143 obtains information obtained through steps S161 to S166 (acquired from the integration function and integration function of each small region). The number of peaks, peak position, peak ratio, first threshold value, peak ratio image, etc.) are stored in the storage unit 150 via the control unit 130 as feature quantities of each small region. When the process proceeds to step S168 through steps S161 to S163 and S167, the feature amount acquisition unit 143 obtains information obtained through steps S161 to S163 and S167 (acquired from the integration function and integration function of each small region). The number of peaks, peak positions, resolved / unresolved determination results, etc.) are stored in the storage unit 150 via the control unit 130 as feature quantities of each small region.

  Then, when the process of step S168 ends, the process of step S160 of FIG. 4 ends.

Here, it returns to description of FIG. 4 again.
When the process of step S160 in FIG. 4 ends, the process proceeds to step S170.

<Step S170>
In step S170, the region extraction unit 144 performs unresolved regions in which the photoreceptor cells are unresolved regions in the AO-SLO image acquired in step S130, based on the feature amounts of the small regions acquired in step S160. Processing for extracting an image region is performed. For example, the region extraction unit 144 performs a process of extracting an unresolved region based on a distribution of peak positions that are feature amounts of each small region set in the AO-SLO image.

  Specifically, in the present embodiment, when the peak ratio image is generated in step S160 (when the processing of S166 is performed), the region extraction unit 144 includes the AO-SLO image acquired in step S130. An unresolved area is extracted. In this case, the region extraction unit 144 sets a second threshold for the peak ratio image, sets a region where the pixel value of the peak ratio image is less than the second threshold as an unresolved region, and sets the pixel value of the peak ratio image to the second threshold. The above area is defined as a resolution area.

  FIG. 8B shows a boundary 810 of the unresolved area when the second threshold is set as 0.9 for the peak ratio image shown in FIG. 8A and an unresolved area is extracted. Here, the inside of the boundary 810 of the unresolved area shown in FIG. 8B corresponds to the unresolved area. The reason why the second threshold is set to 0.9 is that when I (ri_high) and I (ri_low) have the same magnitude (the second threshold is 1.0), the resolution is insufficient by visual observation. It depends on what is felt. The second threshold value is not limited to 0.9 exemplified here, and when an evaluation index other than visual observation exists, an optimal value based on the evaluation index may be selected. Further, in step S170, the region extraction unit 144 acquires the length of the boundary 810 and the area of the unresolved region surrounded by the boundary 810 based on the boundary 810 of the unresolved region shown in FIG.

In step S160 (specifically, S167), when it is determined that the entire area of the AO-SLO image acquired in step S130 is resolved, unresolved areas are included in the AO-SLO image. Since it does not exist, the length and area of the boundary 810 of the unresolved area are zero.
In step S160 (specifically, S167), when it is determined that the entire area of the AO-SLO image acquired in step S130 is unresolved, the boundary 810 of the unresolved area is specified. Therefore, its length and area are indefinite.

  Then, the region extraction unit 144 stores the information obtained by the processing in step S170 (the extracted unresolved region, the boundary length and the area thereof, and the like) via the control unit 130. To remember.

<Step S180>
Subsequently, in step S180, the output unit 160 sends the control unit 130 the information about the AO-SLO image acquired in step S130, the unresolved area acquired in step S170, the boundary length, and the area thereof. Via the storage unit 150. Then, the output unit 160 outputs the acquired information and the like to the display device 400, and performs a process shown to the examiner by displaying an unresolved region superimposed on the AO-SLO image on the display device 400. . At the same time, the output unit 160 outputs the acquired information and the like to the database 500 and stores it in the database 500.

  When the process of step S180 ends, the process of the flowchart shown in FIG. 4 ends.

As described above, in the image processing apparatus 100 according to the first embodiment, the acquired AO-SLO image is divided into small regions, and feature amounts acquired from the spatial frequency images of each small region, more specifically, In this case, the feature amount of the shape of the integrated image corresponding to the arrangement period of the photoreceptor cells is obtained. The image processing apparatus 100 according to the first embodiment extracts an unresolved region in which the photoreceptor cell is an unresolved region in the acquired AO-SLO image based on the feature amount of each small region. I have to.
According to such a configuration, since the unresolved region in which the photoreceptor cell is an unresolved region is extracted from the AO-SLO image, by excluding the unresolved region from the photoreceptor cell detection target region, It becomes possible to accurately detect the photoreceptor cells of the eye to be examined. Specifically, for example, it is possible to prevent an index such as a density obtained as a result of photoreceptor cell analysis from being an uncertain value.

(Second Embodiment)
Next, a second embodiment of the present invention will be described.

  In the first embodiment, the unresolved region near the fovea is extracted based on the feature amount acquired from the spatial frequency image of the small region. At this time, the approximate fovea position is set from the lighting position information of the fixation lamp acquired at the time of shooting, and the distance from the fovea of each image and each small area is the approximate fovea position. Was set based on.

  In the second embodiment, the position of the central fovea of the eye to be examined is acquired from the unresolved area extracted by the area extraction unit 144. Specifically, in the second embodiment, the position of the center of gravity of the unresolved area extracted by the area extraction unit 144 is acquired as the position of the fovea of the eye to be examined. Here, as a method for obtaining the position of the fovea, a plurality of methods are known, such as when referring to the travel of capillaries, but the most common method is to detect the density of photoreceptor cells and the density thereof. A method for obtaining the position having the highest value. However, in this case, the position accuracy of the fovea depends on the accuracy of photoreceptor detection, and accurate position detection is often difficult.

  The schematic configuration of the ophthalmic apparatus according to the second embodiment is the same as the schematic configuration of the ophthalmic apparatus 10 according to the first embodiment shown in FIG.

[Processing Procedure of Image Processing Apparatus 100]
FIG. 9 is a flowchart illustrating an example of a processing procedure of an image processing method by the image processing apparatus 100 according to the second embodiment of the present invention. Here, in FIG. 9, processing steps similar to those shown in FIG. 4 are denoted by the same reference numerals, and detailed description thereof is omitted.

  First, in the image processing method by the image processing apparatus 100 according to the second embodiment, the processes of steps S110 to S170 in FIG. 4 are performed.

<Step S910>
Subsequently, in step S910, the photoreceptor analyzing unit 145 obtains the position of the center of gravity of the unresolved area extracted in step S170, and uses the position of the center of gravity of the unresolved area as the position of the fovea of the eye to be examined. get. Thereby, the position of the fovea with higher accuracy than that of the first embodiment can be acquired. The photoreceptor analysis unit 145 that acquires the position of the fovea constitutes a fovea acquisition unit.

  However, if it is determined in step S160 (specifically, S167) that the entire area of the AO-SLO image acquired in step S130 is unresolved, the position of the fovea cannot be acquired. Indefinite. Further, when it is determined in step S160 (specifically, S167) that the entire area of the AO-SLO image is resolved, the position of the fovea cannot be acquired by the method of this example. Therefore, in this case, the present embodiment uses a method in which the position where the density of photoreceptor cells is the highest is based on the detection result of photoreceptor cells, which is a general technique.

  The photoreceptor analysis unit 145 stores the foveal position information acquired in this manner in the storage unit 150 via the control unit 130.

<Step S920>
Subsequently, in step S920, the output unit 160 displays the information such as the AO-SLO image acquired in step S130, the unresolved area acquired in step S170, the foveal position information acquired in step S910, and the like. Via the storage unit 150. Then, the output unit 160 outputs the acquired information and the like to the display device 400, and the display device 400 displays the unresolved region and the foveal position superimposed on the AO-SLO image. The process shown in is performed. At the same time, the output unit 160 outputs the acquired information and the like to the database 500 and stores it in the database 500.

  When the process of step S920 ends, the process of the flowchart shown in FIG. 9 ends.

According to the second embodiment, the following effects can be obtained in addition to the effects of the first embodiment described above.
That is, even if the density of photoreceptor cells is high near the fovea and areas where the photoreceptor cells are not completely resolved remain, the center of gravity of the unresolved area where the photoreceptor cells are unresolved areas is centered. By obtaining the position of the fovea, it is possible to obtain a more accurate position of the fovea.

(Third embodiment)
Next, a third embodiment of the present invention will be described.

  In the first embodiment, the unresolved region near the fovea is extracted based on the feature amount acquired from the spatial frequency image of the small region. Furthermore, in the second embodiment, a more accurate fovea position is obtained from the extracted unresolved area.

  The third embodiment masks the unresolved area extracted by the area extracting unit 144 in the acquired AO-SLO image in order to perform a more accurate photoreceptor analysis near the fovea of the eye to be examined. Is set, and analysis such as detection of photoreceptor cells is performed for the analysis target region. Furthermore, in the third embodiment, by using the position of the fovea with higher accuracy acquired in the second embodiment, it is possible to further improve the analysis accuracy near the fovea where the density change due to the distance is large. It becomes possible.

  The schematic configuration of the ophthalmologic apparatus according to the third embodiment is the same as the schematic configuration of the ophthalmologic apparatus 10 according to the first embodiment (same as the second embodiment) shown in FIG. To do.

[Processing Procedure of Image Processing Apparatus 100]
FIG. 10 is a flowchart illustrating an example of a processing procedure of an image processing method by the image processing apparatus 100 according to the third embodiment of the present invention. Here, in FIG. 10, the same processing steps as the processing steps shown in FIG. 9 are denoted by the same reference numerals, and detailed description thereof will be omitted.

  First, in the image processing method by the image processing apparatus 100 according to the third embodiment, the processes of steps S110 to S170 and S910 of FIG. 9 are performed.

<Step S1010>
Subsequently, in step S1010, the photoreceptor analysis unit 145 sets an analysis target region by masking the unresolved region extracted in step S170 in the AO-SLO image acquired in step S130, and the analysis target Visual cells are detected for the region. The photoreceptor analysis unit 145 that detects photoreceptor cells constitutes photoreceptor cell detection means.

  Here, a plurality of methods are known as methods for detecting photoreceptor cells. For example, Non-Patent Document 1 shows a method of focusing on the luminance gradient of an AO-SLO image and detecting its maximum value. Has been. However, in order for a photoreceptor cell to be detected in an image as a maximum value, each photoreceptor cell needs to be resolved individually. Therefore, there is a high possibility that the detection result obtained from the unresolved area is not accurately detected. Therefore, in the present embodiment, in view of this point, the unresolved area extracted in step S170 is masked in order to exclude the unresolved area that generates such an inaccurate detection result from the analysis target area. . Specifically, for example, only the detection points included in the analysis target region are set as detection results for the detection results of photoreceptor cells acquired from the entire image.

  Furthermore, when performing Voronoi analysis based on the detection points, only Voronoi regions included in the analysis target region are targeted. Specifically, for example, Voronoi analysis is first performed using detection points from the entire image without considering the mask region. As a result, Voronoi regions for all detection points are acquired. Among them, the Voronoi region's mother point and Voronoi points are selected only from the Voronoi region included in the analysis target region excluding the mask region. The Voronoi region included in.

  The photoreceptor analysis unit 145 stores information on the photoreceptor cell detection results thus detected (for example, photoreceptor cell analysis results such as photoreceptor cell detection points and Voronoi regions) via the controller 130. Store in the unit 150.

<Step S1020>
Subsequently, in step S1020, the output unit 160 displays the AO-SLO image acquired in step S130, the information such as the unresolved area acquired in step S170, the foveal position information acquired in step S910, and the step S1010. The detection result information and the like of the photoreceptor cells acquired in (1) is acquired from the storage unit 150 via the control unit 130. Then, the output unit 160 outputs the acquired information and the like to the display device 400, and the display device 400 superimposes the unresolved region, the position of the fovea, and the detection result of the photoreceptor cell on the AO-SLO image. The processing shown to the examiner is performed by displaying it. At the same time, the output unit 160 outputs the acquired information and the like to the database 500 and stores it in the database 500.

  When the process of step S1020 ends, the process of the flowchart shown in FIG. 10 ends.

According to the third embodiment, in addition to the effects in the first embodiment described above, the following effects are provided.
In other words, even when the density of photoreceptor cells is high near the fovea and areas where the photoreceptor cells are not completely resolved remain, the photoreceptor cells are detected by masking the unresolved area, thereby enabling visual inspection of the eye to be examined. Cells can be detected more accurately.

(Fourth embodiment)
Next, a fourth embodiment of the present invention will be described.

  In the first embodiment, the unresolved region near the fovea is extracted based on the feature amount acquired from the spatial frequency image of the small region. In the second embodiment, the position of the fovea is acquired from the extracted unresolved area. Further, in the third embodiment, the photoreceptor cell is detected by masking the extracted unresolved region.

  However, when an image with high image quality cannot be acquired at the photographing stage, it is difficult to acquire a highly accurate analysis result of photoreceptor cells. Various factors are known for the deterioration of the image quality, and one of them is a focus shift at the time of photographing. The AO-SLO image is subjected to aberration correction at the time of shooting, and is adjusted so that the surface where the magnitude of reflected light is highest is in focus. In principle, it is considered that the visual cell layer is in focus, but in reality, the visual cell layer may not be correctly focused due to the movement of the eye to be examined or the influence from the upper layer of the retina.

  Therefore, in the fourth embodiment, images obtained by shifting the focus position up and down from the focus position set by the aberration correction are acquired, the unresolved area of each image is acquired, and the size is the largest. A method for selecting a smaller focus position will be described.

  The schematic configuration of the ophthalmologic apparatus according to the fourth embodiment is the same as the schematic configuration of the ophthalmologic apparatus 10 according to the first embodiment (the same applies to the second and third embodiments) shown in FIG. Description is omitted.

[Processing Procedure of Image Processing Apparatus 100]
FIG. 11 is a flowchart illustrating an example of a processing procedure of an image processing method performed by the image processing apparatus 100 according to the fourth embodiment of the present invention. Here, in FIG. 11, the same processing steps as the processing steps shown in FIG. 4 are denoted by the same reference numerals, and detailed description thereof is omitted. Specifically, step S110 of FIG. 11 is the same as the process of step S110 in the first embodiment shown in FIG. 11 are substantially the same as steps S120 and S140 to S170 in the first embodiment shown in FIG. 4, but the process is performed on a plurality of images having different focus positions. Only the point is different. Therefore, in the following description, only this difference part will be described.

  First, in the image processing method by the image processing apparatus 100 according to the fourth embodiment, the process of step S110 in FIG. 4 is performed.

<Step S1110>
Subsequently, in step S <b> 1110, the image acquisition unit 110 acquires a plurality of planar images (AO-SLO images or the like) having different focus positions from the AO-SLO apparatus 200. Then, the image acquisition unit 110 stores the acquired plurality of planar images in the storage unit 150 via the control unit 130.

  Here, the plurality of planar images having different focus positions are a group of planar images photographed by moving the focus position up and down around the optimum focus position for the photoreceptor layer acquired by the aberration correction. More specifically, it is assumed that the focus position is moved within a range of about ± 150 μm from the optimum focus position to the choroid and vitreous. In addition, a superimposed image may be acquired by shooting a plurality of times at each focus position, or an image shot while moving the focus position with emphasis on real-time characteristics at the time of shooting. For example, when an image having an S angle of view with a frame rate of 32 is taken for 0.5 seconds (number of images n = 16) while moving from the optimal focus position by 160 μm from the vitreous side by 20 μm to the choroid side, etc. Conceivable.

<Step S1120>
Subsequently, in step S1120, the dividing unit 142 performs a process of dividing each planar image (each AO-SLO image) in the plurality of planar images acquired in step S1110 into small regions. Then, the dividing unit 142 stores information related to the size and position of the small area divided in each planar image (each AO-SLO image) in the storage unit 150 via the control unit 130. However, here, the division method and the size of the small area are the same for each AO-SLO image.

<Step S1130>
Subsequently, in step S1130, the feature amount acquisition unit 143 performs a spatial frequency conversion process on each small region in each AO-SLO image divided in step S1120. Then, the feature amount acquisition unit 143 stores the spatial frequency image Fni of each small region Rni in each AO-SLO image obtained in this way in the storage unit 150 via the control unit 130.

<Step S1140>
Subsequently, in step S1140, the feature amount acquisition unit 143 uses the spatial frequency image Fni of each small region in each AO-SLO image obtained in step S1130 to determine whether the image is resolved or unresolved. Get feature values. The detailed processing procedure of step S1140 is shown in the flowchart shown in FIG. 5 as in the first embodiment. The feature quantity acquisition unit 143 then obtains the feature quantity obtained in this step (the number of peaks based on the integration function of each small region and its shape, the position of the peak, the peak ratio and the peak ratio image, or the entire image) The resolution / unresolved determination result when a plurality of types of peaks are not mixed in the storage unit 150 is stored in the storage unit 150 via the control unit 130.

<Step S1150>
In step S1150, the region extraction unit 144 extracts an unresolved region in each AO-SLO image based on the peak ratio image corresponding to each AO-SLO image acquired in step S1140. Then, the region extraction unit 144 acquires the boundary length of the extracted unresolved region and the area of the unresolved region. If it is determined in step S1140 (specifically, S167) that the entire area of the AO-SLO image acquired in step S1110 is resolved, unresolved areas are included in the AO-SLO image. Since it does not exist, the length and area of the boundary of the unresolved area are zero. In step S160 (specifically, S167), if it is determined that the entire area of the AO-SLO image acquired in step S1110 is unresolved, the length of the boundary of the unresolved area is The peripheral length and the area of the entire AO-SLO image are the area of the entire image.

  Then, the region extracting unit 144 stores the information obtained by the processing in step S1150 (the extracted unresolved region, the boundary length and the area thereof, and the like) via the control unit 130. To remember.

<Step S1160>
Subsequently, in step S1160, for example, the control unit 130 images the eye to be examined by the AO-SLO apparatus 200 based on the area (size) of the unresolved area of each AO-SLO image extracted in step S1150. Determine the focus position when doing it. The control unit 130 that determines the focus position constitutes a focus position determination unit.

  Specifically, for example, the control unit 130 obtains a relationship between the focus position of each AO-SLO image acquired in step S1110 and the area of the unresolved area of each AO-SLO image extracted in step S1150. A focus position that minimizes the area of the unresolved area is determined. Here, a plurality of methods can be considered as a method for determining the focus position. The simplest method is to select an n-AO-SLO image that has the smallest area of the unresolved area, and select and determine the focus position at that time. Alternatively, there is a method in which a graph showing the relationship between the focus position and the area of the unresolved area is created, and the minimum value is calculated and determined using various interpolation methods. As an interpolation method at this time, spline interpolation, quadratic interpolation, or the like can be considered.

  In step S1150, when the area of the unresolved area of all the AO-SLO images has a single value, the focus position is optimal for the photoreceptor layer acquired by the aberration correction in step S1110. Determined as the focus position. In step S1140, such a case occurs when all image areas of all AO-SLO images are resolved or when all image areas of all AO-SLO images are unresolved. .

  For example, the control unit 130 stores the focus position information thus determined in the storage unit 150.

<Step S1170>
Subsequently, in step S1170, the control unit 130 feeds back the focus position determined in step S1160 to the AO-SLO device 200, and performs imaging control for performing re-imaging to acquire an AO-SLO image again at the determined focus position. I do.

  When the process of step S1170 ends, the process of the flowchart shown in FIG. 11 ends.

According to the fourth embodiment, in addition to the effects in the first embodiment described above, the following effects are provided.
That is, by determining and setting the focus position so that the unresolved region of the photoreceptor cell is as small as possible when the density of the photoreceptor cell is high near the fovea and the region where the photoreceptor cell is not completely resolved remains, It is possible to acquire an AO-SLO image that allows easy detection of photoreceptor cells.

  The flowchart shown in FIG. 11 is a mode in which steps S1160 and S1170 are added to the processing steps corresponding to steps S110 to S120 and S140 to S170 shown in FIG. However, the present embodiment is not limited to this aspect. For example, as the present embodiment, for example, a mode in which steps S1160 and S1170 shown in FIG. 11 are added to the processing steps corresponding to steps S110 to S120, S140 to S170 and S910, and S1010 shown in FIG. 10 is also applicable. That is, in the case of this aspect, in addition to the processing of FIG. 11, the foveal position acquisition processing in step S910 and the photoreceptor detection processing in step S1010 shown in FIG. 10 are further performed.

(Other embodiments)
The present invention supplies a program that realizes one or more functions of the above-described embodiments to a system or apparatus via a network or a storage medium, and one or more processors in a computer of the system or apparatus read and execute the program This process can be realized. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.
This program and a computer-readable storage medium storing the program are included in the present invention.

  Note that the above-described embodiments of the present invention are merely examples of implementation in practicing the present invention, and the technical scope of the present invention should not be construed as being limited thereto. It is. That is, the present invention can be implemented in various forms without departing from the technical idea or the main features thereof.

DESCRIPTION OF SYMBOLS 10 Ophthalmology apparatus, 100 Image processing apparatus, 110 Image acquisition part, 120 Information acquisition part, 130 Control part, 140 Image processing part, 141 Position alignment part, 142 Division | segmentation part, 143 Feature-value acquisition part, 144 Area extraction part, 145 views Cell analysis unit, 150 storage unit, 160 output unit, 200 AO-SLO device, 300 information input device, 400 display device, 500 database

Claims (11)

Image obtaining means for obtaining an image obtained by photographing the photoreceptor cells of the eye to be examined;
Dividing means for dividing the image into small regions;
Feature amount acquisition means for performing spatial frequency conversion on each of the small regions divided by the dividing unit and acquiring the feature amount of each of the small regions;
An image processing apparatus comprising: an area extracting unit configured to extract an unresolved area in which the photoreceptor cell is an unresolved area in the image based on the feature amount of each of the small areas.   The image processing apparatus according to claim 1, wherein the feature amount acquisition unit acquires the feature amount based on a shape of an integration function obtained by integrating the absolute values of outputs of the small regions in an angular direction.   The image processing apparatus according to claim 2, wherein the feature amount includes a number of peaks and a peak position in the shape of the integration function.   The said area extraction means extracts the said unresolved area | region based on distribution of the peak position which is the feature-value acquired in each small area | region set in the said image. The image processing apparatus according to claim 1.   When the region extraction means has one type of peak, which is a feature amount acquired in each small region set in the image, the image is resolved according to the type of the peak. The image processing apparatus according to claim 1, further comprising a determination unit that determines whether the image is unresolved.   The image processing apparatus according to claim 1, further comprising a foveal acquisition unit configured to acquire a position of a fovea of the eye to be examined from the unresolved region.   The image processing apparatus according to claim 6, wherein the fovea acquisition unit acquires the position of the center of gravity of the unresolved area as the position of the fovea.   8. The image processing apparatus according to claim 1, further comprising a photoreceptor detection unit that detects the photoreceptor by masking the unresolved area in the image. 9.   The image processing according to any one of claims 1 to 8, further comprising a focus position determination unit that determines a focus position when performing the shooting based on a size of the unresolved area. apparatus. An image acquisition step of acquiring an image of the photoreceptor cells of the eye to be examined;
A dividing step of dividing the image into small regions;
A feature amount acquisition step of performing spatial frequency conversion on each small region divided in the dividing step, and acquiring a feature amount of each small region;
And a region extracting step of extracting an unresolved region in which the photoreceptor cell is an unresolved region based on the feature amount of each small region. An image acquisition step of acquiring an image of the photoreceptor cells of the eye to be examined;
A dividing step of dividing the image into small regions;
A feature amount acquisition step of performing spatial frequency conversion on each small region divided in the dividing step, and acquiring a feature amount of each small region;
A program for causing a computer to execute an area extraction step of extracting an unresolved area in which the photoreceptor cell is an unresolved area in the image based on the feature amount of each small area.