Grade evaluation device, ophthalmic photography device, program, recording medium, and grade evaluation method

Abstract

To quantitatively evaluate a grade of a blood vessel distribution image acquired by OCT angiography on the basis of a plurality of criteria.SOLUTION: A grade evaluation device 500 for a blood vessel distribution image includes: an acquisition unit 510 for acquiring a blood vessel distribution image, which is an image indicating the distribution of blood vessels in the eyeball acquired by OCT angiography; an evaluation value calculation unit 520 for calculating a plurality of evaluation values on the basis of a plurality of criteria from the blood vessel distribution image; and a grade value calculation unit 530 for calculating a grade value, which is a value indicating the grade of the blood vessel distribution image on the basis of the plurality of evaluation values.SELECTED DRAWING: Figure 5

Description

[Technical field]

[0001] The present invention relates to a grade evaluation device, an ophthalmic photography device, a program, a recording medium, and a grade evaluation method. [Background technology]

[0002] Image diagnosis plays an important role in the field of ophthalmology. In recent years, the use of optical coherence tomography (OCT) has been increasing. OCT is now being used not only to obtain B-scan images and 3D images of the examined eye, but also to obtain en-face images such as C-scan images and shadowgrams.

[0003] In addition, OCT can be used to obtain images in which specific parts of the subject's eye are emphasized, and to obtain functional information. For example, OCT angiography has been attracting attention in recent years.

[0004] OCT angiography is a technique that constructs vessel-enhanced images (angiograms, angiograms, motion contrast images) based on time-series data collected by applying OCT to a three-dimensional region of the fundus, typically.

[0005] OCT angiography can obtain an image showing the distribution of blood vessels (hereafter referred to as a blood vessel distribution image). OCT angiography is based on images obtained by OCT. When multiple OCT cross-sectional images are taken at different times at the same location and the cross-sectional images are compared, the signal intensity changes in areas where blood vessels are present due to the movement of red blood cells. This area where the signal intensity changes can be extracted as blood vessels and highlighted to obtain an image of the blood vessels. A blood vessel distribution image obtained by OCT angiography is an image that shows the distribution of blood vessels, for example, in the superficial layer of the retina.

[0006] To obtain a vascular distribution image by OCT angiography, multiple OCT images taken at the same location but at different times are required. However, if tracking does not work well or if the fixation of the examinee's eye is not stable, the quality (grade) of the vascular distribution image obtained by OCT angiography will deteriorate. The grade of the vascular distribution image has traditionally been evaluated visually by doctors. However, it has not been possible to quantitatively evaluate the grade of the vascular distribution image. In addition, although there are multiple possible criteria for evaluating the grade of a vascular distribution image, it has not been possible to quantitatively evaluate the grade of the vascular distribution image based on multiple criteria.

[0007] Patent Document 1 describes an OCT signal processing device that processes an OCT signal detected by an OCT device based on a measurement light scanned over a subject and a reference light. The OCT signal processing device includes a control means for causing a display means to display a confirmation screen for confirming the quality of three-dimensional motion contrast data obtained by processing a plurality of OCT signals from the same site at different times. The control means causes a motion contrast image based on depth region data, which is three-dimensional motion contrast data extracted in a depth region of a portion of the subject, to be displayed on the confirmation screen.

[0008] The technology described in Patent Document 1 displays a confirmation screen on a display means so that a human can check the quality of the three-dimensional motion contrast data, but the device does not quantitatively evaluate the grade of the vascular distribution image.

[0009] Patent Document 2 describes an ophthalmic imaging device that images a subject's eye. The ophthalmic imaging device includes a first acquisition means that acquires a plurality of OCT signals that differ in time for the same position on the subject's eye by an OCT optical system that acquires tomographic image data of the subject's eye. The ophthalmic imaging device includes an image processing means that processes the plurality of OCT signals in the depth direction at each scanning position acquired by the first acquisition means to acquire front motion contrast data or three-dimensional motion contrast data of the subject's eye. The ophthalmic imaging device includes an image division means that divides the front motion contrast data or three-dimensional motion contrast data into a plurality of regions. The ophthalmic imaging device includes a determination means that determines, for each divided region, whether the front motion contrast data or the three-dimensional motion contrast data of each divided region divided by the image division means is appropriate.

[0010] The technology described in Patent Document 2 judges the suitability of the front motion contrast data for each divided region or the three-dimensional motion contrast data for each divided region for each divided region, and reacquires a plurality of OCT signals at each scanning position in the same divided region as the front motion contrast data of the divided region judged to be inappropriate or the three-dimensional motion contrast data of the divided region judged to be inappropriate by the second acquisition means, and sets the reacquired plurality of OCT signals at each scanning position as the plurality of OCT signals in the divided region judged to be inappropriate. That is, the technology described in Patent Document 2 divides the region and reacquires OCT signals for the divided region judged to be inappropriate in consideration of the fact that positional deviations and motion artifacts occur in the image due to the influence of blinking, fixational micromovement, etc. of the subject's eye, thereby acquiring front motion contrast data useful for diagnosis or three-dimensional motion contrast data useful for diagnosis. The technology described in Patent Document 2 does not quantitatively evaluate the grade of the vascular distribution image itself based on a plurality of criteria. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] JP 2017-006179 A [Patent Document 2] JP 2017-046976 A Summary of the Invention [Problem to be solved by the invention]

[0012] An object of the present invention is to provide a grade assessment device, an ophthalmic photography device, a program, a recording medium, and a grade assessment method that are capable of quantitatively assessing the grade of a vascular distribution image obtained by OCT angiography based on multiple criteria. [Means for solving the problem]

[0013] The present disclosure provides an imaging system including an imaging device including: an acquisition unit that acquires a blood vessel distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography; an evaluation value calculation unit that calculates a plurality of evaluation values ​​based on a plurality of criteria from the blood vessel distribution image; and a grade value calculation unit that calculates a grade value that indicates the grade of the blood vessel distribution image based on the plurality of evaluation values. the plurality of evaluation values ​​include an evaluation value indicating the number of horizontal line artifacts in the blood vessel distribution image, an evaluation value indicating the number of black bands in the blood vessel distribution image, an evaluation value indicating an S / N ratio when blood vessels in the blood vessel distribution image are regarded as signals and anything other than blood vessels is regarded as noise, an evaluation value indicating the clarity of blood vessels in the blood vessel distribution image, an evaluation value indicating the degree of connectivity of blood vessels in the blood vessel distribution image, an evaluation value indicating noise in the horizontal line direction in the blood vessel distribution image, and an evaluation value indicating a degree of contrast variation in the blood vessel distribution image. A vascular distribution image grade evaluation device is provided.

[0014] The present disclosure provides an ophthalmic imaging device capable of performing optical coherence tomography (OCT) that includes the above-mentioned grading device.

[0015] The present disclosure provides a program for causing a computer to function as an acquisition unit that acquires a vascular distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography, an evaluation value calculation unit that calculates a plurality of evaluation values ​​based on a plurality of criteria from the vascular distribution image, and a grade calculation unit that calculates a value indicating the grade of the vascular distribution image based on the plurality of evaluation values. wherein the plurality of evaluation values ​​include an evaluation value indicating the number of horizontal line artifacts in the blood vessel distribution image, an evaluation value indicating the number of black bands in the blood vessel distribution image, an evaluation value indicating an S / N ratio when blood vessels in the blood vessel distribution image are regarded as signals and anything other than blood vessels is regarded as noise, an evaluation value indicating the clarity of blood vessels in the blood vessel distribution image, an evaluation value indicating the degree of connectivity of blood vessels in the blood vessel distribution image, an evaluation value indicating noise in a horizontal line direction in the blood vessel distribution image, and an evaluation value indicating a degree of contrast variation in the blood vessel distribution image. Offer the program.

[0016] The present disclosure provides a computer-readable non-transitory recording medium having the above-mentioned program recorded thereon.

[0017] The present disclosure relates to a method for evaluating the grade of a vascular distribution image by a computer having a processor, the method including: a step of calculating, by the processor, a plurality of evaluation values ​​based on a plurality of criteria from a vascular distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography; and a step of calculating, by the processor, a grade value, which is a value indicating the grade of the vascular distribution image, based on the plurality of evaluation values. the plurality of evaluation values ​​include an evaluation value indicating the number of horizontal line artifacts in the blood vessel distribution image, an evaluation value indicating the number of black bands in the blood vessel distribution image, an evaluation value indicating an S / N ratio when blood vessels in the blood vessel distribution image are regarded as signals and anything other than blood vessels is regarded as noise, an evaluation value indicating the clarity of blood vessels in the blood vessel distribution image, an evaluation value indicating the degree of connectivity of blood vessels in the blood vessel distribution image, an evaluation value indicating noise in the horizontal line direction in the blood vessel distribution image, and an evaluation value indicating a degree of contrast variation in the blood vessel distribution image. A grading method is provided. Effect of the Invention

[0018] According to the present disclosure, it is possible to provide a grade assessment device, an ophthalmic imaging device, a program, a recording medium, and a grade assessment method that are capable of quantitatively assessing the grade of a vascular distribution image obtained by OCT angiography based on multiple criteria. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 is a schematic diagram illustrating an example of the configuration of an ophthalmic imaging apparatus. [Diagram 2] FIG. 1 is a schematic diagram illustrating an example of the configuration of an ophthalmic imaging apparatus. [Diagram 3] FIG. 1 is a schematic diagram illustrating an example of the configuration of an ophthalmic imaging apparatus. [Figure 4A] FIG. 1 is a schematic diagram illustrating an example of a scan pattern that can be performed by an ophthalmic imaging device. [Figure 4B] FIG. 1 is a schematic diagram illustrating an example of a scan pattern that can be performed by an ophthalmic imaging device. [Diagram 5] FIG. 1 is a schematic diagram illustrating an example of a configuration of a grade evaluation device according to an embodiment. [Figure 6] A flowchart showing an example of the operation of the grade evaluation device according to the embodiment. [Figure 7] FIG. 13 is a diagram illustrating an example of a vascular distribution image when the grade of the vascular distribution image is Poor. [Figure 8]FIG. 13 is a diagram illustrating an example of a vascular distribution image in which the grade of the vascular distribution image is Average. [Figure 9] FIG. 13 is a diagram illustrating an example of a vascular distribution image in which the grade of the vascular distribution image is Good. [Figure 10] Schematic diagram of the first algorithm [Figure 11] Schematic diagram of the first algorithm [Figure 12] Schematic diagram of the first algorithm [Figure 13] Schematic diagram of the second algorithm [Figure 14] Schematic diagram of the second algorithm [Figure 15] Schematic diagram of the second algorithm [Figure 16] Schematic diagram of the third algorithm [Figure 17] Schematic diagram of the third algorithm [Figure 18] Schematic diagram of the third algorithm [Figure 19] Schematic diagram of the third algorithm [Figure 20] Schematic diagram of the fourth algorithm [Figure 21] Schematic diagram of the fourth algorithm [Figure 22] Schematic diagram of the fifth algorithm [Diagram 23] Schematic diagram of the fifth algorithm [Figure 24] Schematic diagram of the fifth algorithm [Diagram 25] Schematic diagram of the fifth algorithm [Figure 26] Schematic diagram of the sixth algorithm [Figure 27] Schematic diagram of the sixth algorithm [Figure 28] Schematic diagram of the sixth algorithm [Figure 29] Schematic diagram of the sixth algorithm [Diagram 30] Schematic diagram of the sixth algorithm [Diagram 31] Schematic diagram of the seventh algorithm [Diagram 32] Schematic diagram of the seventh algorithm [Diagram 33] Schematic diagram of the 8th algorithm [Diagram 34] FIG. 1 is a conceptual diagram illustrating a decision tree used by a grade evaluation device according to an embodiment; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Several embodiments of the present invention will be described in detail with reference to the accompanying drawings. An ophthalmic imaging apparatus related to the embodiments is an ophthalmic apparatus having at least a function of performing optical coherence tomography (OCT).

[0021] Hereinafter, an ophthalmic imaging device combining a swept source OCT and a fundus camera will be described, but the ophthalmic imaging device is not limited to this. For example, the type of OCT is not limited to swept source OCT, and may be spectral domain OCT or the like. Here, swept source OCT is a method of splitting light from a wavelength-variable light source (wavelength swept light source) into measurement light and reference light, superimposing return light of the measurement light from the test object on the reference light to generate interference light, detecting this interference light with a balanced photodiode or the like, and forming an image by applying Fourier transform or the like to detection data collected according to the wavelength sweep and the measurement light scan. On the other hand, spectral domain OCT is a method of splitting light from a low-coherence light source into measurement light and reference light, superimposing return light of the measurement light from the test object on the reference light to generate interference light, detecting the spectral distribution of this interference light with a spectroscope, and forming an image by applying Fourier transform or the like to the detected spectral distribution. In other words, swept-source OCT is a method for acquiring a spectral distribution in a time division manner, and spectral-domain OCT is a method for acquiring a spectral distribution in a space division manner. Note that OCT methods applicable to the embodiment are not limited to these, and any other method (e.g., time-domain OCT) can be applied.

[0022] The ophthalmologic imaging device may or may not have a function of acquiring a photograph (digital photograph) of the subject's eye, such as a fundus camera. In addition, instead of a fundus camera, any modality, such as a scanning laser ophthalmoscope (SLO), a slit lamp microscope, an anterior segment imaging camera, or a surgical microscope, may be provided. Note that a front image such as a fundus photograph can be used for observing the fundus, setting a scan area, tracking, and the like.

[0023] <Configuration Example of Ophthalmic Imaging Device> As shown in Fig. 1, the ophthalmologic imaging apparatus 1 includes a fundus camera unit 2, an OCT unit 100, and an arithmetic and control unit 200. The fundus camera unit 2 is provided with an optical system similar to that of a conventional fundus camera. The OCT unit 100 is provided with an optical system and mechanisms for performing OCT. The arithmetic and control unit 200 includes a processor. A chin rest and a forehead rest for supporting the face of the subject are provided in positions facing the fundus camera unit 2.

[0024] In this specification, the term "processor" refers to a circuit such as a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), a programmable logic device (e.g., a simple programmable logic device (SPLD), a complex programmable logic device (CPLD), a field programmable gate array (FPGA)), etc. The processor realizes the functions according to the embodiment by, for example, reading and executing a program stored in a memory circuit or a storage device.

[0025] Fundus Camera Unit 2 The fundus camera unit 2 is provided with an optical system and mechanism for photographing the fundus Ef of the subject's eye E. Images obtained by photographing the fundus Ef (called fundus images, fundus photographs, etc.) include observation images and photographed images. Observation images are obtained, for example, by capturing video using near-infrared light. Photographed images are, for example, color images or monochrome images obtained using visible flash light, or monochrome images obtained using near-infrared flash light. The fundus camera unit 2 may be capable of acquiring fluorescein fluorescent images, indocyanine green fluorescent images, spontaneous fluorescent images, etc.

[0026] The fundus camera unit 2 includes an illumination optical system 10 and an imaging optical system 30. The illumination optical system 10 irradiates illumination light onto the subject's eye E. The imaging optical system 30 detects return light of the illumination light from the subject's eye E. The measurement light from the OCT unit 100 is guided to the subject's eye E through an optical path in the fundus camera unit 2, and the return light is guided to the OCT unit 100 through the same optical path.

[0027] The observation light source 11 of the illumination optical system 10 is, for example, a halogen lamp or a light-emitting diode (LED). The light (observation illumination light) output from the observation light source 11 is reflected by a reflecting mirror 12 having a curved reflecting surface, passes through a condenser lens 13, and passes through a visible cut filter 14 to become near-infrared light. The observation illumination light is then focused once in the vicinity of the imaging light source 15, reflected by a mirror 16, and passes through relay lenses 17 and 18, an aperture 19, and a relay lens 20. The observation illumination light is then reflected by the peripheral portion (the area around the hole) of the aperture mirror 21, passes through a dichroic mirror 46, and is refracted by the objective lens 22 to illuminate the subject's eye E (particularly the fundus Ef).

[0028] The return light of the observation illumination light from the subject's eye E is refracted by the objective lens 22, passes through the dichroic mirror 46, passes through the hole formed in the central region of the aperture mirror 21, passes through the dichroic mirror 55, passes through the photographing focusing lens 31, and is reflected by the mirror 32. Furthermore, this return light passes through the half mirror 33A, is reflected by the dichroic mirror 33, and is imaged on the light receiving surface of the CCD image sensor 35 by the condenser lens 34. The CCD image sensor 35 detects the return light at, for example, a predetermined frame rate. When the photographing optical system 30 is focused on the fundus Ef, an observation image of the fundus Ef is obtained, and when the focus is on the anterior segment, an observation image of the anterior segment is obtained.

[0029] The imaging light source 15 is, for example, a visible light source including a xenon lamp or an LED. Light (imaging illumination light) output from the imaging light source 15 is irradiated onto the fundus Ef through the same path as the observation illumination light. Return light of the imaging illumination light from the subject's eye E is guided to the dichroic mirror 33 through the same path as the return light of the observation illumination light, passes through the dichroic mirror 33, is reflected by a mirror 36, and is imaged on the light receiving surface of a CCD image sensor 38 by a condenser lens 37.

[0030] The liquid crystal display (LCD) 39 displays a fixation target for fixating the subject's eye E. A part of the light beam (fixation light beam) output from the LCD 39 is reflected by the half mirror 33A, reflected by the mirror 32, passes through the photographing focusing lens 31 and the dichroic mirror 55, and passes through the hole of the aperture mirror 21. The fixation light beam that passes through the hole of the aperture mirror 21 passes through the dichroic mirror 46, is refracted by the objective lens 22, and is projected onto the fundus Ef. The fixation position of the subject's eye E can be changed by changing the display position of the fixation target on the screen of the LCD 39. Instead of the LCD 39, a matrix LED in which a plurality of LEDs are arranged two-dimensionally, a combination of a light source and a variable diaphragm (such as a liquid crystal diaphragm), or the like can be used as a fixation light beam generating means.

[0031] The fundus camera unit 2 is provided with an alignment optical system 50 and a focus optical system 60. The alignment optical system 50 generates an alignment index used for alignment of the optical system with respect to the subject's eye E. The focus optical system 60 generates a split index used for focus adjustment with respect to the subject's eye E.

[0032] The alignment light output from the LED 51 of the alignment optical system 50 passes through apertures 52 and 53 and a relay lens 54, is reflected by a dichroic mirror 55, and passes through a hole in the aperture mirror 21. The light that has passed through the hole in the aperture mirror 21 passes through the dichroic mirror 46 and is projected onto the subject's eye E by the objective lens 22.

[0033] The corneal reflected light of the alignment light passes through the objective lens 22, dichroic mirror 46 and the hole, part of which passes through the dichroic mirror 55, passes through the photographing focusing lens 31, is reflected by the mirror 32, passes through the half mirror 33A, is reflected by the dichroic mirror 33, and is projected by the condenser lens 34 onto the light receiving surface of the CCD image sensor 35. Based on the light receiving image (alignment index image consisting of two bright points) by the CCD image sensor 35, manual alignment or auto alignment can be performed in the same manner as in the conventional method.

[0034] The focus optical system 60 is moved along the optical path (illumination optical path) of the illumination optical system 10 in conjunction with the movement of the photographing focusing lens 31 along the optical path (photographing optical path) of the photographing optical system 30. The reflecting rod 67 is insertable into and removable from the illumination optical path.

[0035] When performing focus adjustment, the reflecting surface of the reflecting rod 67 is obliquely disposed in the illumination light path. The focusing light output from the LED 61 passes through the relay lens 62, is split into two light beams by the split target plate 63, passes through the two-hole diaphragm 64, is reflected by the mirror 65, and is once imaged on the reflecting surface of the reflecting rod 67 by the condenser lens 66 and is then reflected. The focusing light further passes through the relay lens 20, is reflected by the aperture mirror 21, passes through the dichroic mirror 46, is refracted by the objective lens 22, and is projected onto the fundus Ef.

[0036] The fundus reflection of the focusing light passes through the same path as the cornea reflection of the alignment light and is detected by the CCD image sensor 35. Based on the received light image (split target image consisting of two bright line images) by the CCD image sensor 35, manual alignment and auto alignment can be performed in the same way as in the past.

[0037] The photographing optical system 30 includes diopter correction lenses 70 and 71. The diopter correction lenses 70 and 71 can be selectively inserted into the photographing optical path between the aperture mirror 21 and the dichroic mirror 55. The diopter correction lens 70 is a plus (+) lens for correcting strong hyperopia, for example a +20D (diopter) convex lens. The diopter correction lens 71 is a minus (-) lens for correcting strong myopia, for example a -20D concave lens. The diopter correction lenses 70 and 71 are attached to, for example, a turret plate. The turret plate has a hole formed therein for the case where neither of the diopter correction lenses 70 nor 71 is applied.

[0038] The dichroic mirror 46 combines the optical path for fundus imaging with the optical path for OCT. The dichroic mirror 46 reflects light in the wavelength band used for OCT and transmits light for fundus imaging. The optical path for OCT includes, in order from the OCT unit 100 side, a collimator lens unit 40, an optical path length changer 41, an optical scanner 42, an OCT focusing lens 43, a mirror 44, and a relay lens 45.

[0039] The optical path length changing unit 41 is movable in the direction of the arrow shown in Fig. 1 and changes the optical path length of the optical path for OCT. This change in the optical path length is used for correcting the optical path length according to the axial length of the subject's eye E, adjusting the interference state, etc. The optical path length changing unit 41 includes, for example, a corner cube and a mechanism for moving it.

[0040] The optical scanner 42 is disposed at a position optically conjugate with the pupil of the subject's eye E. The optical scanner 42 changes the traveling direction of the measurement light LS passing through the optical path for OCT. This causes the subject's eye E to be scanned with the measurement light LS. The optical scanner 42 can deflect the measurement light LS in any direction on the xy plane, and includes, for example, a galvanometer mirror that deflects the measurement light LS in the x direction and a galvanometer mirror that deflects the measurement light LS in the y direction.

[0041] <OCT Unit 100> As illustrated in Fig. 2, the OCT unit 100 is provided with an optical system for performing OCT of the subject's eye E. The configuration of this optical system is similar to that of conventional swept-source OCT. That is, this optical system includes an interference optical system that splits light from a light source into measurement light and reference light, superimposes return light of the measurement light from the subject's eye E and the reference light that has passed through a reference light path to generate interference light, and detects this interference light. The detection result (detection signal) obtained by the interference optical system is a signal indicating the spectrum of the interference light, and is sent to the arithmetic and control unit 200.

[0042] The light source unit 101 includes a tunable light source capable of changing the wavelength of emitted light at high speed, similar to a typical swept-source OCT. The tunable light source is, for example, a near-infrared laser light source.

[0043] Light L0 output from light source unit 101 is guided by optical fiber 102 to polarization controller 103, where its polarization state is adjusted. Furthermore, light L0 is guided by optical fiber 104 to fiber coupler 105, where it is split into measurement light LS and reference light LR.

[0044] The reference light LR is guided by an optical fiber 110 to a collimator 111 where it is converted into a parallel beam, and is then guided to a corner cube 114 via an optical path length correction member 112 and a dispersion compensation member 113. The optical path length correction member 112 acts to match the optical path length of the reference light LR with the optical path length of the measurement light LS. The dispersion compensation member 113 acts to match the dispersion characteristics between the reference light LR and the measurement light LS.

[0045] The corner cube 114 bends the incident reference light LR in the opposite direction to the traveling direction of the reference light LR. The incident direction and the exit direction of the reference light LR with respect to the corner cube 114 are parallel to each other. The corner cube 114 is movable in the incident direction of the reference light LR, thereby changing the optical path length of the reference light LR.

[0046] 1 and 2, both an optical path length changer 41 for changing the length of the optical path (measurement optical path, measurement arm) of the measurement light LS and a corner cube 114 for changing the length of the optical path (reference optical path, reference arm) of the reference light LR are provided, but only one of the optical path length changer 41 and the corner cube 114 may be provided. It is also possible to change the difference between the measurement optical path length and the reference optical path length by using optical members other than these.

[0047] The reference light LR that has passed through the corner cube 114 passes through the dispersion compensation member 113 and the optical path length correction member 112, is converted from a parallel beam to a focused beam by the collimator 116, and enters the optical fiber 117. The reference light LR that has entered the optical fiber 117 is guided to a polarization controller 118 where its polarization state is adjusted, is guided by the optical fiber 119 to an attenuator 120 where the amount of light is adjusted, and is guided by the optical fiber 121 to a fiber coupler 122.

[0048] On the other hand, the measurement light LS generated by the fiber coupler 105 is guided by the optical fiber 127 and converted into a parallel beam by the collimator lens unit 40, passes through the optical path length changing unit 41, the optical scanner 42, the OCT focusing lens 43, the mirror 44 and the relay lens 45, is reflected by the dichroic mirror 46, is refracted by the objective lens 22 and enters the subject's eye E. The measurement light LS is scattered and reflected at various depth positions in the subject's eye E. The return light of the measurement light LS from the subject's eye E travels in the opposite direction along the same path as the outward path and is guided to the fiber coupler 105, and reaches the fiber coupler 122 via the optical fiber 128.

[0049] The fiber coupler 122 generates interference light by superimposing the measurement light LS incident via the optical fiber 128 and the reference light LR incident via the optical fiber 121. The fiber coupler 122 generates a pair of interference light LC by splitting the interference light at a predetermined splitting ratio (for example, 1:1). The pair of interference light LC is guided to a detector 125 via optical fibers 123 and 124, respectively.

[0050] The detector 125 is, for example, a balanced photodiode. The balanced photodiode has a pair of photodetectors that detect a pair of interference lights LC, respectively, and outputs the difference between the detection results by these. The detector 125 sends the detection result (detection signal) to a data acquisition system (DAQ) 130.

[0051] The DAQ130 is supplied with a clock KC from the light source unit 101. The clock KC is generated in the light source unit 101 in synchronization with the output timing of each wavelength swept within a predetermined wavelength range by the wavelength-tunable light source. For example, the light source unit 101 optically delays one of two branched lights obtained by branching the light L0 of each output wavelength, and generates the clock KC based on the result of detecting the combined light. The DAQ130 samples the detection signal input from the detector 125 based on the clock KC. The DAQ130 sends the sampling result of the detection signal from the detector 125 to the arithmetic and control unit 200.

[0052] <Arithmetic control unit 200> The arithmetic and control unit 200 controls each part of the fundus camera unit 2, the display device 3, and the OCT unit 100. The arithmetic and control unit 200 also executes various arithmetic processing. For example, the arithmetic and control unit 200 performs signal processing such as Fourier transform on the spectral distribution based on the detection result obtained by the detector 125 for each series of wavelength scans (for each A-line), thereby forming a reflection intensity profile for each A-line. Furthermore, the arithmetic and control unit 200 forms image data by imaging the reflection intensity profile of each A-line. The arithmetic and control unit 200 performs the arithmetic processing for this purpose in the same manner as in conventional swept-source OCT.

[0053] The arithmetic control unit 200 includes, for example, a processor, a RAM (Random Access Memory), a ROM (Read Only Memory), a hard disk drive, a communication interface, etc. Various computer programs are stored in a storage device such as a hard disk drive. The arithmetic control unit 200 may also include an operation device, an input device, a display device, etc.

[0054] <Control System> FIG. 3 shows an example of the configuration of a control system (processing system) of the ophthalmic photographing apparatus 1.

[0055] <Control unit 210> The control unit 210 controls each unit of the ophthalmologic imaging apparatus 1. The control unit 210 includes a processor, RAM, ROM, a hard disk drive, etc. The functions of the control unit 210 are realized by cooperation between hardware including circuits and control software. The control unit 210 includes a main control unit 211 and a storage unit 212.

[0056] <Main control unit 211> The main control unit 211 performs various controls. For example, the main control unit 211 controls the photographing focusing lens 31, the CCD image sensor 35, the CCD image sensor 38, the LCD 39, the optical path length change unit 41, the optical scanner 42, the OCT focusing lens 43, the focus optical system 60, the reflecting rod 67, the light source unit 101, the reference driving unit 114A, the detector 125, the DAQ 130, and the like. The reference driving unit 114A moves the corner cube 114 provided in the reference optical path. This changes the length of the reference optical path.

[0057] <Storage section 212> The storage unit 212 stores various types of data. Examples of data stored in the storage unit 212 include image data of an OCT image, image data of a fundus image, and information about the subject's eye. The information about the subject's eye includes information about the subject, such as a patient ID and name, identification information for the left eye / right eye, and electronic medical record information.

[0058] Data Processing Unit 230 The data processing unit 230 executes various data processing. For example, the data processing unit 230 forms image data based on data collected by the DAQ 130. The data processing unit 230 can also apply processing (signal processing) to the data collected by the DAQ 130, apply image processing or analysis processing to image data formed from data collected by the DAQ 130, and apply image processing or analysis processing to image data output from the CCD image sensor 35 (or the CCD image sensor 38). The data processing unit 230 includes, for example, at least one of a processor and a dedicated circuit board. The data processing unit 230 includes an image forming unit 231.

[0059] <Image forming unit 231> The image forming unit 231 includes an image forming processor (not shown). The image forming unit 231 forms cross-sectional image data of the fundus oculi Ef based on the sampling result of the detection signal input from the DAQ 130 to the data processing unit 230. This processing includes signal processing such as noise removal (noise reduction), filtering, and fast Fourier transform (FFT), as in conventional swept-source OCT. The image data formed by the image forming unit 231 is a data set including a group of image data (a group of A-scan image data) formed by imaging the reflection intensity profile in a plurality of A-lines (scan lines along the z-direction) arranged along a scan line.

[0060] When OCT angiography is performed, the image forming unit 231 can form a motion contrast image based on detection data (e.g., a group of detection signals from the DAQ 130) collected by scanning repeated a predetermined number of times. This motion contrast image is an angiogram in which the blood vessels of the fundus Ef are emphasized. Note that a motion contrast image is an image created based on multiple data (images) acquired at different times at the same position, and is an image that represents the motion at that position.

[0061] Here, a typical example of a scan pattern applicable to OCT angiography will be described. In OCT angiography, a three-dimensional scan (raster scan) is applied. The three-dimensional scan is a scan along a plurality of scan lines arranged parallel to each other. The plurality of scan lines are pre-ordered, and the scan is applied in this order. An example of a three-dimensional scan applicable in this embodiment is shown in FIG. 4A and FIG. 4B.

[0062] As shown in FIG. 4B, the three-dimensional scan in this example is performed on 320 scan lines L1 to L320. One scan along one scan line Li (i=1 to 320) is called a B scan. One B scan consists of 320 A scans (see FIG. 4A). An A scan is a scan for one A line. In other words, an A scan is a scan for an A line along the incident direction (z direction, depth direction, axial direction) of the measurement light LS. A B scan consists of 320 A scans arranged along a scan line Li on an xy plane perpendicular to the z direction.

[0063] In the three-dimensional scan of this example, B scans for scan lines L1 to L320 are performed four times each in any order. The four B scans for each scan line Li are called repetition scans. The order of the four repetitions for each scan line Li is arbitrary. For example, the four scans may be performed consecutively, or B scans for other scan lines may be performed between the four scans.

[0064] The scan lines L1 to L320 are classified into groups of five according to their arrangement order. Each of the 64 groups obtained by this classification is called a unit, and the scan for each unit is called a unit scan. A unit scan consists of four B-scans (repetitions) for each of the five scan lines. In other words, a unit scan consists of 20 B-scans.

[0065] The image forming unit 231 classifies the data collected by applying such a scan pattern into a data set (time series data) for each scan line Li. Here, the data set includes four B-scan data corresponding to four repetitions. Each of the four B-scan data is data collected by one B-scan for the scan line Li.

[0066] Furthermore, the image forming unit 231 forms a motion contrast image corresponding to each scan line Li based on the data set corresponding to the scan line Li. The motion contrast image corresponding to each scan line Li is a two-dimensional angiography image representing a B-scan plane (longitudinal cross section) including the scan line Li.

[0067] The process of forming a motion contrast image is performed in the same manner as in conventional OCT angiography. As described above, in this example, the data set corresponding to the scan line Li includes four B-scan data. Each B-scan data is data collected in one B-scan for the scan line Li.

[0068] First, the image forming unit 231 forms a normal OCT image based on each B-scan data. This OCT image is a B-scan image composed of 320 A-scan image data. As a result, four B-scan images corresponding to the scan line Li are obtained.

[0069] Next, the image forming unit 231 identifies an image region that has changed among the four B-scan images. This process includes, for example, a process of calculating the difference between different B-scan images. Each B-scan image is luminance image data (intensity image data) that represents the morphology of the fundus Ef, and image regions that correspond to areas other than blood vessels are considered to be substantially unchanged. On the other hand, considering that backscattering that contributes to the interference signal changes randomly due to blood flow, it can be estimated that an image region in which a change has occurred among the four B-scan images (for example, a pixel whose difference is not zero, or a pixel whose difference is equal to or greater than a predetermined threshold) is a blood vessel region.

[0070] The image forming unit 231 assigns a predetermined pixel value to the pixels in the identified vascular region. This pixel value may be, for example, a relatively high luminance value (expressed bright and white when displayed) or a pseudocolor value. Note that, as with other conventional techniques, it is also possible to identify the vascular region using Doppler OCT or image processing.

[0071] By such processing, 320 two-dimensional angiographic images corresponding to the 320 scan lines L1 to L320 are obtained. The image forming unit 231 arranges the 320 two-dimensional angiographic images according to the arrangement of the 320 scan lines L1 to L320. This processing includes, for example, a process of arranging (embedding) the 320 two-dimensional angiographic images in a single three-dimensional coordinate system according to the arrangement order and arrangement interval (spacing) of the 320 scan lines L1 to L320. That is, stack data of 320 two-dimensional angiographic images according to the arrangement of the 320 scan lines L1 to L320 can be formed. This stack data is an example of an image (three-dimensional angiographic image) representing the three-dimensional distribution of blood vessels of the fundus Ef. The image forming unit 231 can also form volume data (voxel data) by performing an interpolation process or the like on this stack data.

[0072] The process of forming an angiographic image from the collected data is not limited to the above examples, and any known technique can be used to form an angiographic image.

[0073] The image forming unit 231 can process three-dimensional images such as volume data and stack data. For example, the image forming unit 231 can apply rendering to the three-dimensional image. Rendering methods include volume rendering, maximum intensity projection (MIP), minimum intensity projection (MIP), surface rendering, and multiplanar reconstruction (MPR). In addition, the image forming unit 231 can construct projection data and a shadowgram by projecting at least a part of the three-dimensional image in the z direction (A-line direction, depth direction).

[0074] The image forming unit 231 can execute any analysis processing or image processing. For example, the image forming unit 231 can apply segmentation to a two-dimensional cross-sectional image or a three-dimensional image. Segmentation is a process for identifying a partial region in an image. In this example, an image region corresponding to a specific tissue of the fundus Ef can be identified.

[0075] In OCT angiography, the image forming unit 231 can construct any two-dimensional angiographic image and / or any pseudo three-dimensional angiographic image from a three-dimensional angiographic image. For example, the data processing unit 230 can construct a two-dimensional angiographic image representing any cross section of the fundus Ef by applying multiplanar reconstruction to the three-dimensional angiographic image.

[0076] Furthermore, the image forming unit 231 can apply segmentation to the three-dimensional angiographic image to specify an image region corresponding to a specific tissue of the fundus Ef, and construct a shadowgram (frontal angiographic image) by projecting the specified image region in the z-direction. Examples of frontal angiographic images include frontal images corresponding to any depth region of the fundus Ef (e.g., the shallow retina, the deep retina, the choriocapillaris, the sclera, etc.) and frontal images corresponding to specific tissues of the fundus Ef (e.g., the internal limiting membrane, the nerve fiber layer, the ganglion cell layer, the inner plexiform layer, the inner nuclear layer, the outer plexiform layer, the outer nuclear layer, the external limiting membrane, the retinal pigment epithelium, the Bruch's membrane, the choroid, the choroid-scleral boundary, the sclera, any part of these, a combination of at least two or more of these, etc.).

[0077] The image forming unit 231 may include, for example, an image processing processor and an image analysis processor in addition to the image formation processor described above. The image formation processor is realized by the cooperation of hardware including a circuit and image formation software. The image processing processor is realized by the cooperation of hardware including a circuit and image processing software. The image analysis processor is realized by the cooperation of hardware including a circuit and image analysis software. In this specification, "image data" and an "image" based on the image data may be regarded as the same. Also, a part of the subject's eye E may be regarded as the same as an image representing the part.

[0078] Grade evaluation device 500 FIG. 5 is a schematic diagram illustrating an example of the configuration of a grade evaluation device 500 according to the embodiment.

[0079] The grade evaluation device 500 includes a processor 501 and a memory 502. The processor 501 is configured using, for example, a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), or a programmable logic device (for example, a simple programmable logic device (SPLD), a complex programmable logic device (CPLD), or a field programmable gate array (FPGA)), and performs various processes and controls in cooperation with the memory 502. Specifically, the processor 501 refers to a grade evaluation program P stored in the memory 502 and executes the program to functionally realize an acquisition unit 510, an evaluation value calculation unit 520, and a grade value calculation unit 530. The grade evaluation program P may be recorded in, for example, various recording media other than the memory 502.

[0080] The memory 502 includes, for example, a RAM (Random Access Memory) as a work memory used when executing the processing of the processor 501, and a ROM (Read Only Memory) that stores a program that defines the processing of the processor 501. The RAM temporarily stores data generated or acquired by the processor 501. The ROM has written therein a program that defines the processing of the processor 501.

[0081] The memory 502 may store a blood vessel distribution image. The blood vessel distribution image is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography. The blood vessel distribution image may be an image generated or stored by, for example, the control unit 210, the OCT unit 100, or the data processing unit 230 shown in FIG. 3.

[0082] The acquiring unit 510 acquires a vascular distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography. The vascular distribution image acquired by the acquiring unit 510 may be an image stored in the memory 502. The vascular distribution image acquired by the acquiring unit 510 may be an image acquired from an external device viewed from the grade evaluation device 500. When acquiring a vascular distribution image from an external device, the grade evaluation device 500 may include a communication interface for connecting to the external device or a communication network such as the Internet connected to the external device.

[0083] The evaluation value calculation unit 520 calculates a plurality of evaluation values ​​based on a plurality of criteria from the blood vessel distribution image acquired by the acquisition unit 510.

[0084] In this embodiment, a plurality of algorithms are implemented in the evaluation value calculation unit 520. For example, a first algorithm 521, a second algorithm 522, a third algorithm 523, a fourth algorithm 524, a fifth algorithm 525, a sixth algorithm 526, a seventh algorithm 527, and an eighth algorithm 528 are implemented in the evaluation value calculation unit 520. The evaluation value calculation unit 520 calculates a plurality of evaluation values ​​using each of these algorithms. Details of the algorithms will be described later. Note that the algorithms implemented in the evaluation value calculation unit 520 are not limited to the above-mentioned eight algorithms.

[0085] The grade value calculation unit 530 calculates a grade value indicating the grade of the blood vessel distribution image based on the multiple evaluation values ​​calculated by the evaluation value calculation unit 520. Details of the grade value calculated by the grade value calculation unit 530 will be described later.

[0086] The grade evaluation device 500 may further include an input unit for receiving information input by a user, an output unit for outputting information to a user, etc. The grade evaluation device 500 may be integrated with, for example, the control unit 210, the OCT unit 100, the data processing unit 230, etc. shown in FIG.

[0087] The grade assessment device 500 may be incorporated in the ophthalmic imaging apparatus 1. That is, the ophthalmic imaging apparatus 1 capable of performing optical coherence tomography (OCT) may include the grade assessment device 500.

[0088] FIG. 6 is a flowchart showing an example of the operation of the grade assessment device 500 according to the embodiment.

[0089] The acquisition unit 510 acquires a blood vessel distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography (St101).

[0090] The evaluation value calculation unit 520 calculates a plurality of evaluation values ​​based on a plurality of criteria from the blood vessel distribution image acquired in step St101 (St102).

[0091] The grade value calculation unit 530 calculates a grade value indicating the grade of the blood vessel distribution image based on the multiple evaluation values ​​calculated in step St102 (St103).

[0092] 7 is a diagram illustrating a blood vessel distribution image in which the grade of the blood vessel distribution image is Poor (grade value = 3). As shown in the figure, the blood vessel distribution image in which the grade value is 3 contains many horizontal line artifacts and noise, making it difficult for an ophthalmologist or a diagnostic system to make a diagnosis based on the image.

[0093] FIG. 8 is a diagram illustrating a vascular distribution image when the grade of the vascular distribution image is Average (grade value = 2). FIG. 9 is a diagram illustrating a vascular distribution image when the grade of the vascular distribution image is Good (grade value = 1). As shown in the figure, a vascular distribution image when the grade value is 1 or 2 has fewer horizontal line artifacts and noise, etc., compared to a vascular distribution image when the grade value is 3. A vascular distribution image when the grade value is 1 or 2 is suitable for an ophthalmologist or a diagnostic system to make a diagnosis based on the image.

[0094] Note that, in the above example, the vascular distribution image is evaluated using three grades: 3 (Poor), 2 (Average), and 1 (Good). However, the grade of the vascular distribution image is not limited to the above three grades, and may be graded, for example, into two or five grades.

[0095] There are multiple possible criteria for evaluating the grade of a blood vessel distribution image. However, it is difficult for a person such as an ophthalmologist to quantitatively determine the grade of a blood vessel distribution image while always taking into account the multiple criteria. The grade evaluation device 500 according to an embodiment of the present disclosure quantitatively evaluates the grade of a blood vessel distribution image based on the multiple criteria.

[0096] <First algorithm: EyeMotion> Fig. 10 is a schematic diagram of the first algorithm 521. Fig. 11 is a schematic diagram of the first algorithm 521. Fig. 12 is a schematic diagram of the first algorithm 521. The first algorithm 521 executed by the evaluation value calculation unit 520 will be described with reference to Figs. 10 to 12.

[0097] The first algorithm 521 is an algorithm for calculating the number of artifacts generated in the blood vessel distribution image due to eye movement. FIG. 10 shows a blood vessel distribution image (original image) acquired by the acquisition unit 510. The Y-axis direction (vertical direction) and the X-axis direction (horizontal direction) in the image are also shown. In the original image, the bright areas (white areas) are blood vessel regions, and the bright areas (gray or black areas) are regions other than blood vessels (background regions). In the blood vessel distribution image, as seen in the region REG1, horizontal line artifacts may occur in which blood vessels are displaced horizontally due to eye movement. The first algorithm 521 is an algorithm for calculating the number of these horizontal line artifacts as a first evaluation value.

[0098] As shown in FIG. 11, the line HL1 and the line HL2 extend in the horizontal direction in the vascular distribution image. In the portion in the vascular distribution image where the horizontal line artifact occurs, the pixels are shifted in the horizontal direction, so the correlation between the adjacent lines HL1 and HL2 is low. Therefore, the evaluation value calculation unit 520 calculates a value indicating the correlation between the adjacent lines HL1 and HL2. If the value indicating the correlation is lower than a predetermined value, the evaluation value calculation unit 520 determines that a horizontal line artifact exists in the portion. The predetermined value is, for example, 0.3. Then, the evaluation value calculation unit 520 counts the number of horizontal line artifacts in the vascular distribution image and calculates the number of horizontal line artifacts as a first evaluation value. In FIG. 12, the horizontal line artifacts detected by the above method are represented by black lines. The evaluation value calculation unit 520 uses the above first algorithm 521 to calculate 124, which is the number of horizontal line artifacts, as the first evaluation value. Note that the smaller the value of the first evaluation value, the better the grade of the vascular distribution image.

[0099] Second algorithm: BlackBand Fig. 13 is a schematic diagram of the second algorithm 522. Fig. 14 is a schematic diagram of the second algorithm 522. Fig. 15 is a schematic diagram of the second algorithm 522. The second algorithm 522 executed by the evaluation value calculation unit 520 will be described with reference to Figs. 13 to 15.

[0100] 13 shows a blood vessel distribution image (original image) acquired by the acquisition unit 510. The Y-axis direction (vertical direction) and the X-axis direction (horizontal direction) of the image are also shown. In the original image, high-luminance parts (white parts) are blood vessel regions, and low-luminance parts (gray or black parts) are regions other than blood vessels (background regions). In the blood vessel distribution image, black bands may appear instead of blood vessels, as seen in region REG2. The second algorithm 522 is an algorithm for calculating the number of these black bands.

[0101] As shown in FIG. 14, the lines HL1 and HL2 extend in the horizontal direction in the blood vessel distribution image. In a portion where a black band occurs in the blood vessel distribution image, pixels having similar pixel values ​​are continuous across multiple lines. The evaluation value calculation unit 520 calculates a value indicating the correlation between the adjacent lines HL1 and HL2. If the value indicating the correlation is higher than a predetermined value, the evaluation value calculation unit 520 determines that a black band exists in the portion. The predetermined value is, for example, 0.9. The evaluation value calculation unit 520 then counts the black bands in the blood vessel distribution image and calculates the number of black bands as a second evaluation value. For convenience of illustration, the black bands detected by the above method are displayed as white bands in FIG. 15. The smaller the value of the second evaluation value, the better the grade of the blood vessel distribution image.

[0102] <Third algorithm: SNR> Fig. 16 is a schematic diagram of the third algorithm 523. Fig. 17 is a schematic diagram of the third algorithm 523. Fig. 18 is a schematic diagram of the third algorithm 523. Fig. 19 is a schematic diagram of the third algorithm 523. The third algorithm 523 executed by the evaluation value calculation unit 520 will be described with reference to Figs. 16 to 19.

[0103] The third algorithm 523 is an algorithm for evaluating the contrast difference between a vascular region and a non-vascular region in a vascular distribution image. A signal to noise ratio (SNR) is used as a value expressing the contrast difference. In the calculation of the SNR by the evaluation value calculation unit 520, the vascular region is treated as a signal, and the non-vascular region is treated as a noise.

[0104] To calculate the SNR, a person skilled in the art may refer to the descriptions in, for example, the following papers: "Methods and algorithms for optical coherence tomography-based angiography: a review and comparison", A. Zhang et al., Journal of Biomedical Optics 20(10), 100901 (October 2015) "Image quality metrics for optical coherence angiography", A. Lozzi et al., 1 Jul 2015 | Vol. 6, No. 7 | BIOMEDICAL OPTICS EXPRESS 2435-2477

[0105] FIG. 16 shows a blood vessel distribution image (original image) acquired by the acquisition unit 510. The Y-axis direction (vertical direction) and X-axis direction (horizontal direction) of the image are also shown. In the original image, high-luminance parts (white parts) are blood vessel regions, and low-luminance parts (gray or black parts) are regions other than blood vessels (background regions). From this original image, the evaluation value calculation unit 520 generates an image corresponding to the blood vessel region and an image corresponding to the region other than blood vessels. To generate these two types of images, the evaluation value calculation unit 520 performs image processing on the original image to create a mask image.

[0106] 17 shows a mask image generated based on an original image by the evaluation value calculation unit 520. To generate the mask image, the evaluation value calculation unit 520 performs various image processing on the original image, such as adaptive binarization and deleting objects smaller than a certain size.

[0107] Next, the evaluation value calculation unit 520 performs multiplication between the original image and the mask image, thereby obtaining a signal image as shown in Fig. 18, in which the blood vessel region in the original image is extracted.

[0108] On the other hand, the evaluation value calculation unit 520 generates an inverted mask image in which values ​​of the binarized mask image (see FIG. 17) are inverted. The inverted mask image is an image in which black and white in the mask image shown in FIG. 17 are swapped. The evaluation value calculation unit 520 multiplies the original image by the inverted mask image. As a result, a noise image as shown in FIG. 19 is obtained in which areas other than blood vessels in the original image are extracted. The noise image is also called a background image.

[0109] Next, the evaluation value calculation unit 520 calculates the SNR based on the signal image and the noise image in accordance with the following formula.

[0110]

number

[0111] μ s means the average pixel value of the signal image. σ s means the standard deviation of the pixel values ​​of the signal image. b means the average pixel value of the noise image. σ b means the standard deviation of pixel values ​​of the noise image.

[0112] The evaluation value calculation section 520 calculates the SNR as the third evaluation value using the above-mentioned third algorithm 523. Note that the larger the third evaluation value, the better the grade of the blood vessel distribution image.

[0113] 4th Algorithm: Edge Sharpness Fig. 20 is a schematic diagram of the fourth algorithm 524. Fig. 21 is a schematic diagram of the fourth algorithm 524. The fourth algorithm 524 executed by the evaluation value calculation section 520 will be described with reference to Figs. 20 and 21.

[0114] The fourth algorithm 524 is an algorithm for evaluating the clarity of blood vessels appearing in the blood vessel distribution image. In Fig. 20, two blood vessel distribution images (original images) acquired by the acquisition unit 510 are superimposed.

[0115] The evaluation value calculation unit 520 copies the original image acquired by the acquisition unit 510 to generate two original images. For ease of explanation, the first original image is referred to as original image A. The second original image is referred to as original image B. The Y-axis direction (vertical direction) and X-axis direction (horizontal direction) of the images are also shown. In the two original images, the brighter areas (white areas) are blood vessel regions, and the brighter areas (gray or black areas) are regions other than blood vessels (background regions).

[0116] The evaluation value calculation unit 520 shifts the original image B by a predetermined distance along the X-axis and Y-axis of the image. The evaluation value calculation unit 520 overlaps the original image A and the shifted original image B as shown in FIG. 20. Next, the evaluation value calculation unit 520 calculates the absolute difference in pixel values ​​for the area where the two original images A and B overlap. For example, if the pixel value at the coordinates (p, q) of the original image A is a and the pixel value at the coordinates (p, q) of the shifted original image B is b, the evaluation value calculation unit 520 calculates the value of |ab| for the coordinates (p, q). An image defined by the absolute difference (hereinafter referred to as a difference image) is shown in FIG. 21.

[0117] The evaluation value calculation unit 520 calculates the average value for each pixel included in the difference image. The larger the average value, the higher the clarity of the blood vessels. That is, the evaluation value calculation unit 520 uses the average pixel value of the difference image as the fourth evaluation value. Note that the larger the fourth evaluation value, the better the grade of the blood vessel distribution image.

[0118] 5th Algorithm: Connectivity Fig. 22 is a schematic diagram of the fifth algorithm 525. Fig. 23 is a schematic diagram of the fifth algorithm 525. Fig. 24 is a schematic diagram of the fifth algorithm 525. Fig. 25 is a schematic diagram of the fifth algorithm 525. The fifth algorithm 525 executed by the evaluation value calculation unit 520 will be described with reference to Figs. 22 to 25.

[0119] The fifth algorithm 525 is an algorithm for evaluating the degree of connectivity of blood vessels appearing in a blood vessel distribution image. The degree of connectivity of blood vessels is a value indicating the degree of connectivity of blood vessels appearing in a blood vessel distribution image. When blood vessels appearing in a blood vessel distribution image are finely fragmented, the degree of connectivity of blood vessels is interpreted as low. When blood vessels appearing in a blood vessel distribution image are long and connected, the degree of connectivity of blood vessels is interpreted as high. The degree of connectivity of blood vessels can be calculated as follows, for example, by referring to the paper "Split-spectrum amplitude-decorrelation angiography with optical coherence tomography", Y. Jia et al., 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4710-4725.

[0120] 22 shows a blood vessel distribution image (original image) acquired by acquisition unit 510. The Y-axis direction (vertical direction) and X-axis direction (horizontal direction) of the image are also shown. In the original image, areas with high brightness (white areas) are blood vessel regions, and areas with low brightness (gray or black areas) are regions other than blood vessels (background regions). Evaluation value calculation unit 520 binarizes the original image.

[0121] 23 shows a binarized image obtained by binarizing the original image. The evaluation value calculation unit 520 reduces (skeletonizes) the widths of blood vessel regions (white parts in the figure) appearing in the binarized image to a width of one pixel each, thereby generating a reduced image.

[0122] Fig. 24 shows a reduced image in which the width of the blood vessel region in the binarized image has been reduced. Evaluation value calculation unit 520 removes objects that appear in the reduced image and have a pixel count that is equal to or less than a predetermined threshold. Note that a region in the image from which the object has been removed is treated as a region other than blood vessels (background region). The predetermined threshold is, for example, 30. Fig. 25 shows the image after object removal (removed image).

[0123] The evaluation value calculation unit 520 calculates the percentage of the number of pixels in the vascular region in the post-removal image as the fifth evaluation value. At this point, the region remaining in the image as blood vessels has a certain number of connected pixels or more (e.g., 30 pixels or more). In other words, the higher the percentage of pixels in the vascular region, the more highly connected blood vessels appear in the original image. The higher the value of the fifth evaluation value, the better the grade of the vascular distribution image.

[0124] 〈6th Algorithm: NoiseDetection1〉 Fig. 26 is a schematic diagram of the sixth algorithm 526. Fig. 27 is a schematic diagram of the sixth algorithm 526. Fig. 28 is a schematic diagram of the sixth algorithm 526. Fig. 29 is a schematic diagram of the sixth algorithm 526. Fig. 30 is a schematic diagram of the sixth algorithm 526. The sixth algorithm 526 executed by the evaluation value calculation unit 520 will be described with reference to Figs. 26 to 30.

[0125] The sixth algorithm 526 is an algorithm for detecting horizontal line artifacts occurring as noise in the blood vessel distribution image. Fig. 26 shows the blood vessel distribution image (original image) acquired by the acquisition unit 510. The Y-axis direction (vertical direction) and the X-axis direction (horizontal direction) in the image are also shown. In the original image, the bright areas (white areas) are blood vessel regions, and the bright areas (gray or black areas) are regions other than blood vessels (background regions).

[0126] In order to detect horizontal line artifacts included in the blood vessel distribution image, the evaluation value calculation unit 520 applies a Gabor filter. When the Gabor filter is applied, horizontal noise stands out, so that horizontal line artifacts can be detected. However, when the Gabor filter is directly applied to the original image, blood vessels extending in the horizontal direction are also detected as horizontal line artifacts, as shown in FIG. 30. In order to prevent blood vessels extending in the horizontal direction, which are not actually noise, from being detected as horizontal line artifacts, the evaluation value calculation unit 520 first performs preprocessing on the original image.

[0127] FIG. 27 shows an image (preprocessed image) after preprocessing has been performed on the original image. The preprocessing process is, for example, as follows: The evaluation value calculation unit 520 calculates the pixel average value of the entire original image. The evaluation value calculation unit 520 extracts an area with high signal intensity (blood vessel area) from the original image. This extraction may be performed, for example, by extracting only pixels whose signal intensity is equal to or greater than a predetermined threshold. Next, the evaluation value calculation unit 520 changes the value of each pixel included in the extracted area with high signal intensity to the calculated pixel average value. As a result of the above preprocessing, pixels corresponding to the blood vessel area have a signal intensity equivalent to the pixel average value of the original image, and therefore are not detected as horizontal line artifacts after applying a Gabor filter described later.

[0128] The evaluation value calculation unit 520 applies a Gabor filter to the pre-processed image. The image after applying the Gabor filter (filtered image) is shown in FIG. 28. Here, consider the situation when a person such as an ophthalmologist evaluates the grade of the vascular distribution image. Then, it is considered that noise present in the center part of the vascular distribution image has a greater influence on the grade evaluation than noise present in the outer part of the vascular distribution image. Therefore, the evaluation value calculation unit 520 performs a process to change the weight of the detected noise between the center part and the outer part of the vascular distribution image. More specifically, the evaluation value calculation unit 520 performs multiplication between the filtered image and a Gaussian mask. The Gaussian mask is shown in FIG. 28. The image after applying the Gaussian mask to the filtered image (masked image) is shown in FIG. 29.

[0129] The evaluation value calculation section 520 calculates the noise average value of the entire masked image as a sixth evaluation value. Note that the smaller the sixth evaluation value, the better the grade of the vascular distribution image.

[0130] 〈7th Algorithm: NoiseDetection2〉 Fig. 31 is a schematic diagram of the seventh algorithm 527. Fig. 32 is a schematic diagram of the seventh algorithm 527. The seventh algorithm 527 executed by the evaluation value calculation section 520 will be described with reference to Figs. 31 and 32.

[0131] The seventh algorithm 527 is an algorithm for detecting horizontal line artifacts occurring as noise in the blood vessel distribution image. FIG. 31 shows the blood vessel distribution image (original image) acquired by the acquisition unit 510. The Y-axis direction (vertical direction) and the X-axis direction (horizontal direction) of the image are also shown. In the original image, the bright areas (white areas) are blood vessel regions, and the bright areas (gray or black areas) are regions other than blood vessels (background regions). The evaluation value calculation unit 520 performs a fast Fourier transform (FFT) on the original image. The image after the fast Fourier transform (FFT image) is shown in FIG. 32.

[0132] If the original image contains many horizontal line artifacts, a vertical white line appears in the center of the FFT image. Therefore, the evaluation value calculation unit 520 adds up pixel values ​​contained in a predetermined region REG3 in the FFT image in the vertical direction (Y-axis direction) and generates a one-dimensional vector in which multiple added values ​​are arranged in the horizontal direction (X-axis direction). Here, the region REG3 includes the center part of the FFT image in the horizontal direction (X-axis direction). In addition, the region REG3 has a predetermined width. The region REG3 may be the entire region of the FFT image.

[0133] The evaluation value calculation unit 520 calculates the standard deviation SD of the multiple sum values ​​included in the generated one-dimensional vector. When a vertical white line appears in the center of the FFT image, the signal intensity in the center part of the one-dimensional vector is high and the signal intensity in the end part is low. If the original image contains many horizontal line artifacts, the value of the standard deviation SD of the multiple sum values ​​included in the one-dimensional vector becomes large. The evaluation value calculation unit 520 calculates the value of the standard deviation SD as the seventh evaluation value using the seventh algorithm 527. Note that the smaller the value of the seventh evaluation value, the better the grade of the vascular distribution image.

[0134] Here, the evaluation value calculation unit 520 may calculate 1 / SD, which is the reciprocal of the value of the standard deviation SD, as the seventh evaluation value by using the seventh algorithm 527. In this case, the larger the value of the seventh evaluation value, the better the grade of the vascular distribution image.

[0135] 8th Algorithm: LocalContrast Fig. 33 is a schematic diagram of the eighth algorithm 528. The eighth algorithm 528 is an algorithm for evaluating the local contrast of a blood vessel distribution image, that is, the degree of contrast variation. Fig. 33 shows a blood vessel distribution image (original image) acquired by the acquisition unit 510. The Y-axis direction (vertical direction) and the X-axis direction (horizontal direction) of the image are also shown. In the original image, high brightness parts (white parts) are blood vessel regions, and low brightness parts (gray or black parts) are regions other than blood vessels (background regions).

[0136] The evaluation value calculation unit 520 divides the original image into a plurality of sub-images. In the illustrated example, the evaluation value calculation unit 520 divides the original image into N sections in the vertical direction (Y-axis direction) and M sections in the horizontal direction (X-axis direction) to generate N×M sub-images.

[0137] Next, the evaluation value calculation unit 520 calculates the pixel average value for each sub-image. In the illustrated example, the evaluation value calculation unit 520 calculates the pixel average value of N×M pixels. When there is local contrast in the blood vessel distribution image, there is variation among the pixel average values ​​of the N×M pixels. Therefore, the evaluation value calculation unit 520 calculates the standard deviation of the pixel average values ​​of the N×M pixels as an eighth evaluation value. Note that the smaller the eighth evaluation value, the better the grade of the blood vessel distribution image.

[0138] Grading of vascular distribution images based on multiple evaluation values As described above, the evaluation value calculation unit 520 calculates a plurality of evaluation values ​​based on a plurality of criteria. Note that one criterion corresponds to one of the above-mentioned algorithms. Next, the grade value calculation unit 530 calculates a grade value of the vascular distribution image based on the plurality of evaluation values ​​calculated in step St102 (St103). The grade value calculation unit 530 can calculate the grade value of the vascular distribution image by using a regression or a decision tree.

[0139] Grading of vascular images based on regression First, the grade determination based on regression will be described. In this embodiment, a general logistic regression is exemplified as the regression mode. The general formula of the logistic regression is as follows.

[0140]

number

[0141] In the above general formula, y is the objective variable of the logistic regression. The grade value calculation unit 530 may calculate the objective variable y as a grade value. n is a natural number of 2 or more. Also, n is equal to or less than the number of standards (number of algorithms) that the evaluation value calculation unit 520 can use. As shown in FIG. 5, when eight algorithms from the first algorithm 521 to the eighth algorithm 528 are implemented in the evaluation value calculation unit 520, n is equal to or less than 8.

[0142] x1 to x n are explanatory variables for logistic regression. b0 is the error term, and b1 to b n is a partial regression coefficient. Values ​​calculated in advance are used for the error term and the partial regression coefficient. For example, the values ​​of the error term and the partial regression coefficient may be stored in the memory 502 of the grade evaluation device 500 shown in FIG. 5. The grade value calculation unit 530 may also obtain the values ​​of the error term and the partial regression coefficient from an external device.

[0143] When i is a natural number between 1 and n, the grade value calculation unit 530 calculates the explanatory variable x i As an example, when the number of explanatory variables is five, the evaluation value calculation unit 520 may input a value based on the seventh evaluation value into explanatory variable x1, a value based on the sixth evaluation value into explanatory variable x2, a value based on the first evaluation value into explanatory variable x3, a value based on the third evaluation value into explanatory variable x4, and a value based on the eighth evaluation value into explanatory variable x5. The above-mentioned input examples of values ​​are merely examples, and a person skilled in the art may appropriately determine which evaluation value to input into each explanatory variable.

[0144] The value based on the evaluation value may be the evaluation value itself, or may be a value obtained by performing a predetermined calculation on the evaluation value, such as the absolute value, reciprocal, logarithm, or square root of the evaluation value.

[0145] For example, as described above, the grade value calculation unit 530 uses the explanatory variable x iThe value of the objective variable y is calculated using a value based on the evaluation value calculated by the evaluation value calculation unit 520 as above. This y value corresponds to the grade value calculated by the grade value calculation unit 530. As is clear from the above general formula (2), the y value in the case of logistic regression is a continuous value from 0 to 1. In this case, the grade of the vascular distribution image is, for example, Grade 1: Good when the value of y is greater than a predetermined first threshold value; Grade 3: Poor when the value of y is less than a predetermined second threshold value; and Grade 2: Average when the value of y is between the predetermined first threshold value and second threshold value. Note that the first threshold value is greater than the second threshold value.

[0146] In the above, the case where the explanatory variables are five is exemplified. However, the number of explanatory variables may be other than five, for example, eight. In addition, the type of regression used by the grade value calculation unit 530 is not limited to the logistic regression described above, and other types of regression may be used.

[0147] Grading of vascular distribution images based on decision trees Next, the grade determination based on the decision tree will be described. In the grade assessment device 500 according to the embodiment, the grade value calculation unit 530 uses a decision tree. The decision tree may be, for example, a classification tree that classifies the blood vessel distribution image into three classes. The three classes are Grade 1: Good, Grade 2: Average, and Grade 3: Poor.

[0148] 34 is a conceptual diagram illustrating a decision tree 600 used by the grade assessment device 500 according to the embodiment. The first layer of the decision tree 600 has a node 611 which is a root node. Edges extend from the node 611 to nodes 621 and 622 in the second layer. Edges extend from the node 622 to nodes 631 and 632 in the third layer. An edge extends from the node 631 to node 641 in the fourth layer. An edge extends from the node 632 to node 642 in the fourth layer.

[0149] A condition determination using explanatory variables is performed at each node in the decision tree 600. The grade value calculation unit 530 calculates a grade value using the decision tree 600 using the values ​​based on the multiple evaluation values ​​calculated by the evaluation value calculation unit 520 as explanatory variables.

[0150] The grade value calculation unit 530 starts processing from node 611, which is the root node. In node 611, the grade value calculation unit 530 uses the value of 1 / SD, which is the seventh evaluation value related to the seventh algorithm (NoiseDetection2), to branch the subsequent processing to node 621 or node 622. For example, the grade value calculation unit 530 performs the processing of node 621 when the value of 1 / SD is less than a predetermined value, and performs the processing of node 622 when the value of 1 / SD is equal to or greater than the predetermined value.

[0151] In the node 621, the grade value calculation unit 530 uses the SNR value, which is the third evaluation value related to the third algorithm (SNR). For example, when the SNR value is less than a predetermined value, the grade value calculation unit 530 calculates a grade value of 3 (Poor). When the SNR value is equal to or greater than a predetermined value, the grade value calculation unit 530 calculates a grade value of 2 (Average).

[0152] In node 622, the grade value calculation unit 530 uses the value of 1 / SD, which is the seventh evaluation value related to the seventh algorithm (NoiseDetection2), to branch the subsequent processing to node 631 or node 632. For example, the grade value calculation unit 530 performs the processing of node 631 when the value of 1 / SD is less than a predetermined value, and performs the processing of node 632 when the value of 1 / SD is equal to or greater than the predetermined value.

[0153] In node 631, the grade value calculation unit 530 uses the value of SNR, which is a third evaluation value related to the third algorithm (SNR). For example, when the value of SNR is less than a predetermined value, the grade value calculation unit 530 performs the process of node 641. When the value of SNR is equal to or greater than a predetermined value, the grade value calculation unit 530 calculates a grade value of 2 (Average).

[0154] In node 632, the grade value calculation unit 530 uses the noise average value of the entire masked image, which is the sixth evaluation value related to the sixth algorithm (NoiseDetection1). For example, if the noise average value is less than a predetermined value, the grade value calculation unit 530 calculates a grade value of 1 (Good). If the noise average value is equal to or greater than the predetermined value, the grade value calculation unit 530 performs the process of node 642.

[0155] In node 641, the grade value calculation unit 530 uses the standard deviation of the average values ​​of N×M pixels, which is the eighth evaluation value related to the eighth algorithm (LocalContrast). If the standard deviation of the average values ​​of N×M pixels is less than a predetermined value, the grade value calculation unit 530 calculates a grade value of 2 (Average). If the standard deviation of the average values ​​of N×M pixels is equal to or greater than a predetermined value, the grade value calculation unit 530 calculates a grade value of 3 (Poor).

[0156] At node 642, the grade value calculation unit 530 uses the number of horizontal line artifacts as the first evaluation value related to the first algorithm (EyeMotion). For example, if the number of horizontal line artifacts is less than a predetermined value, the grade value calculation unit 530 calculates a grade value of 1 (Good). If the number of horizontal line artifacts is equal to or greater than a predetermined value, the grade value calculation unit 530 calculates a grade value of 2 (Average).

[0157] For example, as described above, the grade value calculation unit 530 of the grade assessment device 500 according to the embodiment of the present disclosure calculates a grade value by the decision tree 600 using values ​​based on a plurality of assessment values ​​as explanatory variables. The decision tree 600 is a classification tree that classifies a blood vessel distribution image into three classes: Good, Average, and Poor. The grade value calculated by the grade value calculation unit 530 takes three types of values ​​(1, 2, and 3) corresponding to the three classes, respectively.

[0158] The structure of the nodes and edges in the decision tree 600 may be determined in advance. For example, information defining the structure of the nodes and edges in the decision tree 600 may be stored in the memory 502 of the grade assessment device 500.

[0159] Note that the structure of the nodes and edges in the decision tree 600 is not limited to that shown in Fig. 26. For example, a person skilled in the art can appropriately change the depth of the hierarchy of the decision tree, a value based on which evaluation value is used as an explanatory variable in each node, the branching condition from each node to the next node (such as a threshold value for comparison with a value based on an evaluation value), the number of classes to be classified, and the like.

[0160] Furthermore, the structure of the nodes and edges in the decision tree may be obtained by machine learning using learning data in which a plurality of evaluation values ​​are explanatory variables and the grade value is a response variable. As an algorithm for performing the machine learning, decision tree learning, which is a conventional technique, may be used.

[0161] As described above, the grade evaluation device 500 for a blood vessel distribution image according to the embodiment includes an acquisition unit 510, an evaluation value calculation unit 520, and a grade value calculation unit 530. The acquisition unit 510 acquires a blood vessel distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography. The evaluation value calculation unit 520 calculates a plurality of evaluation values ​​based on a plurality of criteria from the blood vessel distribution image. The grade value calculation unit 530 calculates a grade value, which is a value indicating the grade of the blood vessel distribution image, based on the plurality of evaluation values. This allows the grade evaluation device 500 to quantitatively evaluate the grade of the blood vessel distribution image obtained by OCT angiography based on a plurality of criteria.

[0162] The grade value calculation unit 530 calculates the grade value by logistic regression using values ​​based on multiple evaluation values ​​as explanatory variables. This allows the grade value calculation unit 530 to calculate the grade value as a continuous value. Therefore, by appropriately adjusting the threshold value for distinguishing the grades, highly accurate grade evaluation can be achieved.

[0163] The grade value calculation unit 530 calculates the grade value by a decision tree using values ​​based on multiple evaluation values ​​as explanatory variables, so that the grade value calculation unit 530 can flexibly use multiple evaluation values ​​to calculate an appropriate grade value.

[0164] The decision tree is a classification tree that classifies the vascular distribution image into three classes, and the grade value has three types of values ​​corresponding to the three classes, respectively. This allows the grade evaluation device 500 to evaluate the grade of the vascular distribution image into three stages: Poor, Average, and Good.

[0165] The structure of nodes and edges in the decision tree is obtained by machine learning using learning data in which a plurality of evaluation values ​​are explanatory variables and the grade value is a target variable. This allows the decision tree to be improved by machine learning. Therefore, the grade evaluation device 500 can perform more accurate grade evaluation.

[0166] The evaluation value calculation unit 520 calculates an evaluation value indicating the number of horizontal line artifacts in the blood vessel distribution image, which enables the grade assessment device 500 to assess the grade of the blood vessel distribution image in consideration of the number of horizontal line artifacts.

[0167] The evaluation value calculation unit 520 calculates an evaluation value indicating the number of black bands in the blood vessel distribution image, which enables the grade evaluation device 500 to perform a grade evaluation of the blood vessel distribution image in consideration of the number of black bands.

[0168] The evaluation value calculation unit 520 calculates an evaluation value indicating an S / N ratio when blood vessels in a blood vessel distribution image are regarded as signals and other than blood vessels are regarded as noise. This enables the grade evaluation device 500 to perform a grade evaluation of the blood vessel distribution image taking the S / N ratio into consideration.

[0169] The evaluation value calculation unit 520 calculates an evaluation value indicating the clarity of blood vessels in the blood vessel distribution image, which enables the grade evaluation device 500 to perform a grade evaluation of the blood vessel distribution image in consideration of the clarity of blood vessels in the blood vessel distribution image.

[0170] The evaluation value calculation unit 520 calculates an evaluation value indicating the degree of connectivity of blood vessels in the blood vessel distribution image, which enables the grade evaluation device 500 to perform a grade evaluation of the blood vessel distribution image in consideration of the degree of connectivity of blood vessels in the blood vessel distribution image.

[0171] The evaluation value calculation unit 520 calculates an evaluation value indicating noise in the horizontal direction in the blood vessel distribution image, which enables the grade evaluation device 500 to perform grade evaluation of the blood vessel distribution image while taking into account noise in the horizontal direction in the blood vessel distribution image.

[0172] The evaluation value calculation unit 520 calculates an evaluation value indicating the degree of contrast variation in the blood vessel distribution image, which enables the grade evaluation device 500 to perform a grade evaluation of the blood vessel distribution image in consideration of the degree of contrast variation in the blood vessel distribution image.

[0173] An ophthalmic imaging device 1 capable of performing optical coherence tomography (OCT) is equipped with a grade evaluation device 500. This allows the ophthalmic imaging device 1 to not only photograph the subject's eye, but also perform OCT to generate a blood vessel distribution image, and further quantitatively evaluate the grade of the blood vessel distribution image based on multiple criteria.

[0174] The program P causes the computer 500 to function as an acquisition unit 510, an evaluation value calculation unit 520, and a grade value calculation unit 530. The acquisition unit 510 acquires a vascular distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography. The evaluation value calculation unit 520 calculates a plurality of evaluation values ​​based on a plurality of criteria from the vascular distribution image. The grade value calculation unit 530 calculates a grade value, which is a value indicating the grade of the vascular distribution image, based on the plurality of evaluation values. In addition, the above program P is recorded in a computer-readable non-transitory recording medium (for example, the memory 502). This allows the grade of the vascular distribution image obtained by OCT angiography to be quantitatively evaluated based on a plurality of criteria.

[0175] A computer 500 having a processor 501 executes a grade evaluation method for a blood vessel distribution image. In the grade evaluation method, the processor 501 calculates a plurality of evaluation values ​​based on a plurality of criteria from a blood vessel distribution image, which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography. The processor 501 calculates a grade value, which is a value indicating the grade of the blood vessel distribution image, based on the plurality of evaluation values. This makes it possible to quantitatively evaluate the grade of the blood vessel distribution image obtained by OCT angiography based on the plurality of criteria.

[0176] Although various embodiments have been described above with reference to the drawings, it goes without saying that the present disclosure is not limited to such examples. It is clear that a person skilled in the art can conceive of various modifications, corrections, substitutions, additions, deletions, and equivalents within the scope of the claims, and it is understood that these also naturally belong to the technical scope of the present disclosure. In addition, the components in the various embodiments described above may be arbitrarily combined within the scope of the invention. [Industrial Applicability]

[0177] INDUSTRIAL APPLICABILITY The present disclosure is useful as a grading device, an ophthalmic imaging device, a program, a recording medium, and a grading method that quantitatively evaluate the grade of a vascular distribution image obtained by OCT angiography based on a plurality of criteria. [Explanation of symbols]

[0178] 1 Ophthalmic imaging equipment 2 Fundus camera unit 3 Display device 10 Illumination optical system 11 Observation light source 12 Reflective mirror 13, 34, 37, 66 Condenser lens 14 Visible light cut filter 15 Light source 16, 21, 32, 36, 44, 65 Mirror 17, 18, 20, 45, 54, 62 Relay lenses 22 Objective Lens 30. Photographing Optical System 31 Focusing lens 33, 46, 55 Dichroic mirror 33A Half Mirror 35, 38 CCD image sensor 39 Liquid crystal display 40 Collimator lens unit 41 Optical path length change unit 42 Optical Scanner 43 OCT focusing lens 50 Alignment Optical System 60 Focus Optical System 63 Split Optotype Plate 67 Reflector rod 70, 71 Diopter correction lenses 100 OCT units 101 Light source unit 102, 104, 110, 117, 119, 121 123, 124, 127, 128 Optical fiber 103, 118 Polarization Controller 105, 122 Fiber Coupler 111, 116 Collimator 112 Optical path length correction member 113 Dispersion compensation materials 114 Corner Cube 114A Reference Drive 120 Attenuator 125 Detector 130 Data Acquisition System (DAQ) 200 Calculation and Control Unit 210 Control section 211 Main control unit 212 Storage section 230 Data Processing Unit 231 Image forming section 500 Grade Evaluation Device 501 Processor 502 Memory 510 Acquisition Department 520 Evaluation value calculation unit 530 Grade Value Calculation Department

Claims

1. an acquisition unit that acquires a blood vessel distribution image that is an image representing the distribution of blood vessels in an eyeball obtained by OCT angiography; an evaluation value calculation unit that calculates a plurality of evaluation values ​​based on a plurality of criteria from the blood vessel distribution image; a grade value calculation unit that calculates a grade value indicating a grade of the vascular distribution image based on the plurality of evaluation values; Equipped with The multiple evaluation values ​​include an evaluation value indicating the number of horizontal line artifacts in the vascular distribution image, an evaluation value indicating the number of black bands in the vascular distribution image, an evaluation value indicating an S / N ratio when blood vessels in the vascular distribution image are considered as signals and anything other than blood vessels is considered as noise, an evaluation value indicating the sharpness of blood vessels in the vascular distribution image, an evaluation value indicating the degree of connectivity of blood vessels in the vascular distribution image, an evaluation value indicating noise in the horizontal line direction in the vascular distribution image, and an evaluation value indicating the degree of contrast variation in the vascular distribution image. A device for grading vascular distribution images.

2. the grade value calculation unit calculates the grade value by logistic regression using values ​​based on the plurality of evaluation values ​​as explanatory variables, The grade evaluation device according to claim 1 .

3. the grade value calculation unit calculates the grade value by a decision tree using values ​​based on the plurality of evaluation values ​​as explanatory variables, The grade evaluation device according to claim 1 .

4. the decision tree is a classification tree that classifies the vascular distribution image into three classes; The grade value has three types of values ​​corresponding to the three classes, respectively. The grading device according to claim 3 .

5. The structure of nodes and edges in the decision tree is obtained by machine learning using learning data in which the plurality of evaluation values ​​are explanatory variables and the grade value is a target variable. The grade evaluation device according to claim 3 or 4.

6. An ophthalmic imaging apparatus capable of performing optical coherence tomography (OCT), comprising the grade evaluation apparatus according to any one of claims 1 to 5.

7. Computer, an acquisition unit that acquires a vascular distribution image, which is an image representing the distribution of blood vessels in an eyeball obtained by OCT angiography; an evaluation value calculation unit that calculates a plurality of evaluation values ​​based on a plurality of criteria from the blood vessel distribution image; and a grade value calculation unit that calculates a value indicating a grade of the vascular distribution image based on the plurality of evaluation values; A program that functions as The multiple evaluation values ​​include an evaluation value indicating the number of horizontal line artifacts in the vascular distribution image, an evaluation value indicating the number of black bands in the vascular distribution image, an evaluation value indicating an S / N ratio when blood vessels in the vascular distribution image are considered as signals and anything other than blood vessels is considered as noise, an evaluation value indicating the sharpness of blood vessels in the vascular distribution image, an evaluation value indicating the degree of connectivity of blood vessels in the vascular distribution image, an evaluation value indicating noise in the horizontal line direction in the vascular distribution image, and an evaluation value indicating the degree of contrast variation in the vascular distribution image. program.

8. A non-transitory computer-readable recording medium having the program according to claim 7 recorded thereon.

9. A method for evaluating a grade of a blood vessel distribution image by a computer having a processor, comprising: A step of the processor calculating a plurality of evaluation values ​​based on a plurality of criteria from a blood vessel distribution image which is an image showing the distribution of blood vessels in an eyeball obtained by OCT angiography; and calculating a grade value indicating a grade of the vascular distribution image based on the plurality of evaluation values ​​by the processor. The multiple evaluation values ​​include an evaluation value indicating the number of horizontal line artifacts in the vascular distribution image, an evaluation value indicating the number of black bands in the vascular distribution image, an evaluation value indicating an S / N ratio when blood vessels in the vascular distribution image are considered as signals and anything other than blood vessels is considered as noise, an evaluation value indicating the sharpness of blood vessels in the vascular distribution image, an evaluation value indicating the degree of connectivity of blood vessels in the vascular distribution image, an evaluation value indicating noise in the horizontal line direction in the vascular distribution image, and an evaluation value indicating the degree of contrast variation in the vascular distribution image. Grading method.

Images