JP2022127783A - Blood flow analysis apparatus, ophthalmologic apparatus, blood flow analysis method, and program - Google Patents

Abstract

To provide a novel technique for correctly evaluating the state of a blood vessel of a subject.SOLUTION: A blood flow analysis apparatus comprises an analysis unit and a classification unit. The analysis unit performs time frequency analysis of time series data of a blood flow speed in an eyeground of a subject. The classification unit classifies the subject into one of a plurality of groups including a healthy person group and a sick person group based on the feature amount of analysis data obtained by the analysis unit.SELECTED DRAWING: Figure 7

Description

The present invention relates to a blood flow analysis device, an ophthalmologic device, a blood flow analysis method, and a program.

Arteriosclerosis is one of the causes of lifestyle-related diseases, and is known to cause heart disease and cerebrovascular disease if it progresses without subjective symptoms. Evaluation of vascular conditions is useful for diagnosing and preventing diseases of the circulatory system such as arteriosclerosis.

There are Cardio Ankle Vascular Index (CAVI), Ankle Brachial Index (ABI), and the like as indices for evaluating vascular conditions. By assessing CAVI, it is possible to determine whether arterial stiffness is in the normal range, suspected arteriosclerosis, or borderline between the normal range and suspected arterial stiffening. By evaluating the ABI, it is possible to determine whether the occlusion state of the lower extremity arteries is within the normal range, mild occlusion, or severe occlusion.

On the other hand, in recent years, a method for evaluating the relationship between the dynamics of systemic circulation and retinal blood flow has attracted attention. For example, Patent Literature 1 and Patent Literature 2 disclose a method of performing optical coherence tomography (OCT) on the fundus of a subject to generate blood flow information in the fundus.

JP 2013-184018 A JP 2019-034237 A

According to a survey by the present inventors, when the vascular status of subjects was determined using CAVI, 15.8% of the diseased group was determined to be within the normal range, and 18.2% of the healthy group was determined to be within the normal range. was determined to be a border region. Similarly, when ABI was used to determine the vascular status of subjects, 57.1% of the sick group was found to be within the normal range.

That is, at present, it is difficult to correctly evaluate the vascular condition only with CAVI or ABI, and a new technique for correctly evaluating the vascular condition of a subject is desired.

The present invention has been made in view of such circumstances, and its object is to provide a new technique for correctly evaluating the vascular condition of a subject.

A first aspect of the embodiment includes an analysis unit that performs time-frequency analysis on time-series data of blood flow velocity in the fundus of a subject; a classifying unit that classifies a subject into one of a plurality of groups including a healthy subject group and a diseased subject group.

In a second aspect of the embodiment, in the first aspect, the analysis unit performs wavelet transform on the time-series data in one beat period, and the classification unit performs wavelet transform data obtained by the analysis unit. The subject is classified into one of the plurality of groups based on the scale value and shift value of the mother wavelet at the feature position of .

In the third aspect of the embodiment, in the first aspect, the analysis unit decomposes the time-series data in one beat period into an approximate term and a minute term at a first depth, and divides the m-th (m is 2 or more A multi-resolution analysis is performed by repeating the decomposition of the approximation term at the (m+1)th depth into the approximation term at the (m+1)th depth and the minute term at the nth (n is a natural number) depth, and the classification unit performs The subject is classified into one of the plurality of groups based on the sum of squares of approximate terms at the n-th depth and the sum of squares of minute terms at the n-th depth.

In a fourth aspect of the embodiment, in the first aspect, the analysis unit includes a first analysis unit that performs wavelet transform on the time series data in one beat period, and the time series data in one beat period. The n-th ( n is a natural number); and a second analysis unit that performs multi-resolution analysis that repeats up to the depth. and placing the subject in one of the plurality of groups based on the sum of squares of approximate terms at the n-th depth and the sum of squares of minute terms at the n-th depth obtained by the second analysis unit Classify.

In a fifth aspect of the embodiment, in the second aspect or the fourth aspect, the feature position is a position where the wavelet transform data is maximum.

In the sixth aspect of the embodiment, in any one of the first to fifth aspects, the blood flow velocity is obtained based on a Doppler signal due to blood flow obtained by performing Doppler OCT on the fundus. be.

A seventh aspect of the embodiment is an optical system that performs OCT on a measurement site in the fundus at different timings, and a blood flow that determines a blood flow velocity in the fundus based on a plurality of signals obtained by the optical system. An ophthalmologic apparatus including an information generator and the blood flow analysis apparatus according to the sixth aspect.

An eighth aspect of the embodiment includes an analysis step of performing a time-frequency analysis on the time-series data of the blood flow velocity in the fundus of the subject, and based on the feature amount of the analysis data obtained in the analysis step, the above and a classification step of classifying the subject into one of a plurality of groups including a healthy subject group and a diseased subject group.

In a ninth aspect of the embodiment, in the eighth aspect, the analyzing step performs wavelet transform on the time-series data in one beat period, and the classifying step performs wavelet transform data obtained in the analyzing step. The subject is classified into one of the plurality of groups based on the scale value and shift value of the mother wavelet at the feature position of .

In the tenth aspect of the embodiment, in the eighth aspect, the analyzing step decomposes the time-series data in one beat period into an approximate term and a minute term at a first depth, and divides the m-th (m is 2 or more A multi-resolution analysis is performed by repeating the decomposition of the approximate term at the (m+1)-th depth into the approximate term and the minute term at the (m+1)-th depth up to the n-th (n is a natural number) depth, and the classification step includes: The subject is classified into one of the plurality of groups based on the sum of squares of approximate terms at the n-th depth and the sum of squares of minute terms at the n-th depth.

In the eleventh aspect of the embodiment, in the eighth aspect, the analysis step includes: a first analysis step of performing a wavelet transform on the time-series data in one beat period; The n-th ( n is a natural number), and a second analysis step of performing multi-resolution analysis iteratively to the depth, wherein the classification step includes scale values and shift values of the mother wavelet at the feature positions of the wavelet transform data obtained in the first analysis step. and placing the subject in one of the plurality of groups based on the sum of squares of approximate terms at the n-th depth and the sum of squares of minute terms at the n-th depth obtained in the second analysis step Classify.

In a twelfth aspect of the embodiment, in the ninth aspect or the eleventh aspect, the feature position is a position where the wavelet transform data is maximum.

In the thirteenth aspect of the embodiment, in any one of the eighth to twelfth aspects, the blood flow velocity is obtained based on a Doppler signal due to blood flow obtained by performing Doppler OCT on the fundus. be.

A fourteenth aspect of the embodiment is a program that causes a computer to execute each step of the blood flow analysis method according to any one of the eighth aspect to the thirteenth aspect.

Note that it is possible to arbitrarily combine the configurations according to the plurality of aspects described above.

ADVANTAGE OF THE INVENTION According to this invention, the new technique for correctly evaluating the vascular condition of a subject can be provided.

It is a schematic diagram showing an example of the configuration of the optical system of the ophthalmologic apparatus according to the first embodiment. It is a schematic diagram showing an example of the configuration of the optical system of the ophthalmologic apparatus according to the first embodiment. 1 is a schematic diagram showing an example of a configuration of a processing system of an ophthalmologic apparatus according to a first embodiment; FIG. 1 is a schematic diagram showing an example of a configuration of a processing system of an ophthalmologic apparatus according to a first embodiment; FIG. 1 is a schematic diagram showing an example of a configuration of a processing system of an ophthalmologic apparatus according to a first embodiment; FIG. 1 is a schematic diagram showing an example of a configuration of a processing system of an ophthalmologic apparatus according to a first embodiment; FIG. 1 is a schematic diagram showing an example of a configuration of a processing system of an ophthalmologic apparatus according to a first embodiment; FIG. 1 is a schematic diagram showing an example of a configuration of a processing system of an ophthalmologic apparatus according to a first embodiment; FIG. 4A and 4B are schematic diagrams for explaining the operation of the ophthalmologic apparatus according to the first embodiment; FIG. 4A and 4B are schematic diagrams for explaining the operation of the ophthalmologic apparatus according to the first embodiment; FIG. 4A and 4B are schematic diagrams for explaining the operation of the ophthalmologic apparatus according to the first embodiment; FIG. 4A and 4B are schematic diagrams for explaining the operation of the ophthalmologic apparatus according to the first embodiment; FIG. FIG. 4 is a flow chart showing an example of the operation of the ophthalmologic apparatus according to the first embodiment; FIG. 4 is a flow chart showing an example of the operation of the ophthalmologic apparatus according to the first embodiment; FIG. 4 is a flow chart showing an example of the operation of the ophthalmologic apparatus according to the first embodiment; 4A and 4B are schematic diagrams for explaining the operation of the ophthalmologic apparatus according to the first embodiment; FIG. 4A and 4B are schematic diagrams for explaining the operation of the ophthalmologic apparatus according to the first embodiment; FIG. FIG. 11 is a schematic diagram showing an example of a configuration example of a processing system of an ophthalmologic apparatus according to a second embodiment; FIG. 11 is a schematic diagram for explaining the operation of the ophthalmologic apparatus according to the second embodiment; FIG. 11 is a schematic diagram for explaining the operation of the ophthalmologic apparatus according to the second embodiment; FIG. 11 is a flow chart showing an example of the operation of the ophthalmologic apparatus according to the second embodiment; FIG. 11 is a schematic diagram for explaining the operation of the ophthalmologic apparatus according to the second embodiment; FIG. 11 is a schematic diagram showing an example of a configuration example of a processing system of an ophthalmologic apparatus according to a third embodiment; FIG. 11 is a schematic diagram showing an example of the configuration of a processing system of an ophthalmologic apparatus according to a fourth embodiment;

Exemplary embodiments of a blood flow analysis device, an ophthalmologic device, a blood flow analysis method, and a program according to the present invention will be described in detail with reference to the drawings. It should be noted that the descriptions of the documents cited in this specification and any known techniques can be incorporated into the following embodiments.

The blood flow analysis apparatus according to the embodiment performs time-frequency analysis on time-series data of the blood flow velocity of the subject, and based on the feature amount of the analysis data obtained by performing the time-frequency analysis, Classify examiners into one of a plurality of groups including a healthy subject group and a diseased subject group. In some embodiments, the blood flow is blood flow in the subject's fundus. In some embodiments the time series data is waveform information. Examples of time-frequency analysis include Continuous Wavelet Transform (CWT), Discrete Wavelet Transform (DWT), Short-Time Fourier Transform (STFT), Multiresolution analysis: MRA), etc. In some embodiments, subjects are classified into one of a plurality of groups including a healthy subject group and a diseased subject group based on two or more feature amounts obtained by two or more time-frequency analyses.

The healthy subject group is a group diagnosed as healthy subjects in advance by a doctor or the like. The sick group is a group diagnosed in advance by a doctor or the like as being sick. In some embodiments, the plurality of groups includes a healthy group, a diseased group, and an intermediate range group between healthy and diseased (suspected diseased group).

In some embodiments, blood flow velocity time series data is obtained using optical coherence tomography (OCT). In some embodiments, the functionality of the blood flow analysis device is installed in an ophthalmic device capable of performing OCT on the subject's eye. In some embodiments, the blood flow analyzer is configured to externally receive blood flow velocity time series data obtained by performing OCT.

A blood flow analysis method according to the embodiment is executed by the blood flow analysis device according to the embodiment. A program according to an embodiment causes a processor (computer) to execute each step of a blood flow analysis method. A recording medium according to the embodiment is a computer-readable non-temporary recording medium (storage medium) in which the program according to the embodiment is recorded.

In this specification, the "processor" includes, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), a programmable logic device (e.g., SPLD (Simple Programmable Logic Device (CPLD) Programmable Logic Device), FPGA (Field Programmable Gate Array)) or the like. The processor implements the functions according to the embodiment by, for example, reading and executing a program stored in a storage circuit or storage device.

In the following embodiments, a case will be described in which an ophthalmologic apparatus that performs Fourier domain type OCT implements the functions of the blood flow analysis apparatus according to the embodiments.

Further, although an ophthalmologic apparatus combining a swept-source OCT and a retinal camera will be described below, the configuration according to the embodiment 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. Swept source OCT divides light from a wavelength tunable light source (wavelength swept light source) into measurement light and reference light, and generates interference light by superimposing return light of the measurement light from the object on the reference light, This interference light is detected by a balanced photodiode or the like, and Fourier transform or the like is performed on the detection data collected according to the wavelength sweep and measurement light scanning to form an image. Spectral domain OCT divides light from a low coherence light source into measurement light and reference light, generates interference light by superimposing the return light of the measurement light from the object on the reference light, and generates the interference light spectrum In this method, the distribution is detected by a spectrometer, and the detected spectral distribution is subjected to Fourier transform or the like to form an image. In other words, swept-source OCT is a method of obtaining spectral distributions in time division, and spectral-domain OCT is a method of obtaining spectral distributions in space division. Note that the OCT technique applicable to the embodiment is not limited to these, and any other technique (for example, time domain OCT) can be applied.

The ophthalmologic apparatus according to the embodiment may or may not have a function of acquiring a photograph (digital photograph) of the subject's eye, such as a fundus camera. Also, 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 instead of the fundus camera. A front image such as a photograph of the fundus can be used for observation of the fundus, setting of a scan area, tracking, and the like.

<First Embodiment>
[Constitution]
As shown in FIG. 1 , an ophthalmologic apparatus 1 functioning as a blood flow analysis apparatus includes a fundus camera unit 2 , an OCT unit 100 and an arithmetic control unit 200 . The retinal camera unit 2 is provided with an optical system substantially similar to that of a conventional retinal camera. The OCT unit 100 is provided with an optical system and a mechanism for performing OCT. Arithmetic control unit 200 includes a processor. A chin rest and a forehead rest for supporting the subject's face are provided at a position facing the retinal camera unit 2 .

[Optical system]
(Fundus camera unit 2)
The fundus camera unit 2 is provided with an optical system and a mechanism for photographing the fundus Ef of the eye E to be examined. An image obtained by photographing the fundus oculi Ef (called a fundus image, a fundus photograph, etc.) includes an observed image and a photographed image. Observation images are obtained, for example, by video shooting using near-infrared light. The captured image is, for example, a color image or a monochrome image obtained using visible flash light, or a monochrome image obtained using near-infrared flash light. The fundus camera unit 2 may be capable of acquiring a fluorescein fluorescence image, an indocyanine green fluorescence image, an autofluorescence image, or the like.

The fundus camera unit 2 includes an illumination optical system 10 and an imaging optical system 30 . The illumination optical system 10 irradiates the eye E to be inspected with illumination light. The imaging optical system 30 detects return light of the illumination light from the eye E to be examined. The measurement light from the OCT unit 100 is guided to the subject's eye E through the optical path in the retinal camera unit 2, and its return light is guided to the OCT unit 100 through the same optical path.

The observation light source 11 of the illumination optical system 10 is, for example, a halogen lamp or a light emitting diode (LED). Light (observation illumination light) output from an observation light source 11 is reflected by a reflecting mirror 12 having a curved reflecting surface, passes through a condenser lens 13, passes through a visible light cut filter 14, and becomes near-infrared light. Become. Furthermore, the observation illumination light is once converged near the photographing light source 15 , reflected by the mirror 16 , and passed through the relay lenses 17 and 18 , the diaphragm 19 and the relay lens 20 . Observation illumination light is reflected by the periphery of the perforated mirror 21 (area around the perforation), passes through the dichroic mirror 46, is refracted by the objective lens 22, and passes through the eye E (especially the fundus oculi Ef). Illuminate.

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 , and passes through the aperture formed in the central region of the apertured mirror 21 . The return light that has passed through the hole passes through the dichroic mirror 55 , passes through the imaging focusing lens 31 , and is reflected by the mirror 32 . Further, 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 . The CCD image sensor 35 detects return light at a predetermined frame rate, for example. When the imaging optical system 30 is focused on the fundus oculi Ef, an observed image of the fundus oculi Ef is obtained, and when the anterior ocular segment is focused on, an observed image of the anterior ocular segment is obtained.

The imaging light source 15 is a visible light source including, for example, a xenon lamp or an LED. The light (imaging illumination light) output from the imaging light source 15 irradiates the fundus oculi Ef through the same path as the observation illumination light. The 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 the mirror 36 , is reflected by the condenser lens 37 . An image is formed on the light receiving surface of the CCD image sensor 38 .

A liquid crystal display (LCD) 39 displays a fixation target for fixing the eye E to be examined. A part of the luminous flux (fixation luminous flux) 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 reaches the hole of the aperture mirror 21. pass through the department. The fixation light flux that has passed through the aperture of the perforated mirror 21 is transmitted through the dichroic mirror 46, refracted by the objective lens 22, and projected onto the fundus oculi Ef. By changing the display position of the fixation target on the screen of the LCD 39, the fixation position of the subject's eye E can be changed. Instead of the LCD 39, a matrix LED in which a plurality of LEDs are two-dimensionally arranged, a combination of a light source and a variable diaphragm (liquid crystal diaphragm, etc.), or the like can be used as fixation light beam generating means.

The retinal 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 target used for alignment of the optical system with respect to the eye E to be examined. The focus optical system 60 generates a split target used for focus adjustment of the eye E to be examined.

Alignment light output from the LED 51 of the alignment optical system 50 passes through the diaphragms 52 and 53 and the relay lens 54 , is reflected by the dichroic mirror 55 , and passes through the aperture of the apertured mirror 21 . The light passing through the hole of the perforated mirror 21 is transmitted through the dichroic mirror 46 and projected onto the subject's eye E by the objective lens 22 .

The cornea-reflected light of the alignment light passes through the objective lens 22 , the dichroic mirror 46 and the hole, and a part of it passes through the dichroic mirror 55 and passes through the imaging focusing lens 31 . The corneal reflected light that has passed through the photographing focusing lens 31 is reflected by the mirror 32, transmitted through the half mirror 33A, reflected by the dichroic mirror 33, and projected onto the light receiving surface of the CCD image sensor 35 by the condenser lens 34. Based on the received image (alignment target image consisting of two bright spots) by the CCD image sensor 35, manual alignment and auto alignment can be performed in the same manner as in the prior art.

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 imaging focusing lens 31 along the optical path (illumination optical path) of the imaging optical system 30 . The reflecting bar 67 can be inserted into and removed from the illumination optical path.

When performing focus adjustment, the reflecting surface of the reflecting bar 67 is obliquely provided in the illumination optical path. The focused light output from the LED 61 passes through a relay lens 62, is split into two light beams by a split optotype plate 63, passes through a two-hole diaphragm 64, is reflected by a mirror 65, and is reflected by a condenser lens 66 onto a reflecting rod. The light is once imaged on the reflecting surface 67 and reflected. Further, the focused light passes through the relay lens 20, is reflected by the perforated mirror 21, passes through the dichroic mirror 46, is refracted by the objective lens 22, and is projected onto the fundus oculi Ef.

The fundus reflected light of the focus light passes through the same path as the corneal reflected light of the alignment light and is detected by the CCD image sensor 35 . Manual alignment and auto-alignment can be performed in the same manner as in the prior art based on the received light image (split target image consisting of two bright line images) by the CCD image sensor 35 .

The dichroic mirror 46 synthesizes the fundus imaging optical path and the OCT optical path. The dichroic mirror 46 reflects light in the wavelength band used for OCT and transmits light for fundus imaging. The optical path for OCT is provided with a collimator lens unit 40, an optical path length changing section 41, an optical scanner 42, an OCT focusing lens 43, a mirror 44, and a relay lens 45 in this order from the OCT unit 100 side.

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 correction of the optical path length according to the axial length of the eye E to be examined, adjustment of the interference state, and the like. The optical path length changer 41 includes, for example, a corner cube and a mechanism for moving it.

The optical scanner 42 is arranged at a position optically conjugate with the pupil of the eye E to be examined. The optical scanner 42 changes the traveling direction of the measurement light LS passing through the optical path for OCT. Thereby, the subject's eye E is 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.

(OCT unit 100)
As illustrated in FIG. 2, the OCT unit 100 is provided with an optical system for performing OCT of the eye E to be examined. The configuration of this optical system is similar to that of conventional swept-source OCT. That is, this optical system divides the light from the light source into measurement light and reference light, and superimposes the return light of the measurement light from the subject's eye E and the reference light that has passed through the reference light path to generate interference light. , includes an interference optical system for detecting this interference light. A 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 control unit 200 .

The light source unit 101 includes a wavelength tunable light source capable of changing the wavelength of emitted light at high speed, like a general swept source OCT. The variable wavelength light source is, for example, a near-infrared laser light source.

The light L0 output from the light source unit 101 is guided to the polarization controller 103 through the optical fiber 102, and its polarization state is adjusted. Further, the light L0 is guided by the optical fiber 104 to the fiber coupler 105 and split into the measurement light LS and the reference light LR.

The reference light LR is guided to the collimator 111 by the optical fiber 110 and converted into a parallel beam, and guided to the corner cube 114 via the optical path length correction member 112 and the dispersion compensation member 113 . The optical path length correction member 112 acts to match the optical path length of the reference light LR and 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.

The corner cube 114 reverses the traveling direction of the incident reference light LR. The direction of incidence and the direction of emission 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.

In the configuration shown in FIGS. 1 and 2, an optical path length changing section 41 for changing the length of the optical path (measurement optical path, measurement arm) of the measurement light LS, and an optical path (reference optical path, reference arm) of the reference light LR. Both corner cubes 114 are provided to change the length. In some embodiments, only one of optical path length changer 41 and corner cube 114 is 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.

The reference light LR that has passed through the corner cube 114 passes through the dispersion compensating member 113 and the optical path length correcting member 112 , is converted by the collimator 116 from a parallel beam to a converged beam, and enters the optical fiber 117 . The reference light LR incident on the optical fiber 117 is guided to the polarization controller 118 to adjust the polarization state, guided to the attenuator 120 by the optical fiber 119 to adjust the light amount, and is sent to the fiber coupler 122 by the optical fiber 121. be guided.

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 light flux by the collimator lens unit 40 . The measurement light LS converted into a parallel light flux 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, and is refracted by the objective lens 22. incident on the eye E to be examined. The measurement light LS is scattered and reflected at various depth positions of the eye E to be examined. 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 forward path, is guided to the fiber coupler 105 , and reaches the fiber coupler 122 via the optical fiber 128 .

The fiber coupler 122 superimposes the measurement light LS input via the optical fiber 128 and the reference light LR input via the optical fiber 121 to generate interference light. The fiber coupler 122 generates a pair of interference lights LC by splitting the interference lights at a predetermined splitting ratio (for example, 1:1). A pair of interference lights LC are guided to detector 125 through optical fibers 123 and 124, respectively.

Detector 125 is, for example, a Balanced Photo Diode. A balanced photodiode has a pair of photodetectors that respectively detect a pair of interference lights LC, and outputs the difference between the detection results of these. The detector 125 sends the detection result (detection signal) to a DAQ (Data Acquisition System) 130 .

A clock KC is supplied from the light source unit 101 to the DAQ 130 . 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. The light source unit 101, for example, optically delays one of the two branched lights obtained by branching the light L0 of each output wavelength, and then outputs the clock KC based on the result of detecting these combined lights. Generate. The DAQ 130 samples the detection signal input from the detector 125 based on the clock KC. DAQ 130 sends the sampling result of the detection signal from detector 125 to arithmetic control unit 200 .

(Arithmetic control unit 200)
The arithmetic control unit 200 controls each part of the retinal camera unit 2 , the display device 3 and the OCT unit 100 . Further, the arithmetic control unit 200 executes various kinds of 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 results obtained by the detector 125 for each series of wavelength scans (for each A line), thereby obtaining forming a reflection intensity profile; Furthermore, the arithmetic and control unit 200 forms image data by imaging the reflection intensity profile of each A line. Arithmetic processing therefor is the same as in conventional swept source OCT.

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

[Processing system]
A processing system (control system) of the ophthalmologic apparatus 1 is configured around an arithmetic control unit 200 .

3 to 8 show configuration examples of the processing system of the ophthalmologic apparatus 1. FIG. FIG. 3 shows an example of a functional block diagram of the processing system of the ophthalmologic apparatus 1. As shown in FIG. FIG. 4 shows an example of a functional block diagram of the image forming section 220 of FIG. FIG. 5 represents an example of a functional block diagram of the data processing unit 300 of FIG. FIG. 6 shows an example of a functional block diagram of the blood flow information generator 320 of FIG. FIG. 7 shows an example of a functional block diagram of the blood flow information analysis section 330 of FIG. FIG. 8 represents an example of a functional block diagram of the analysis unit 350 of FIG.

The arithmetic control unit 200 includes a control section 210, an image forming section 220, and a data processing section 300, as shown in FIG.

(control unit 210)
The control section 210 controls each section of the ophthalmologic apparatus 1 . Control unit 210 includes a processor, RAM, ROM, hard disk drive, and the like. The functions of the control unit 210 are realized by cooperation between hardware including circuits and control software. Control unit 210 includes main control unit 211 and storage unit 212 .

(Main control unit 211)
A main control unit 211 performs various controls. In particular, the main control unit 211 controls the focus driving units 31A and 43A of the retinal camera unit 2, the CCD image sensors 35 and 38, the LCD 39, the optical path length changing unit 41, the optical scanner 42, and the like. The main controller 211 also controls the light source unit 101 of the OCT unit 100, the reference driver 114A, the detector 125, the DAQ 130, and the like. Further, the main controller 211 controls an optical system driver (not shown) that drives the optical system shown in FIGS.

The focus driving section 31A receives control from the main control section 211 and moves the photographing focusing lens 31 along the optical axis of the photographing optical system 30 . The focus driving section 31A is provided with a holding member that holds the photographing focusing lens 31, an actuator that generates driving force for moving the holding member, and a transmission mechanism that transmits this driving force. The actuator is composed of, for example, a pulse motor. The transmission mechanism is configured by, for example, a combination of gears, a rack and pinion, or the like. As a result, the focus driving section 31A controlled by the main control section 211 moves the photographing focusing lens 31, thereby changing the focus position of the photographing optical system 30. FIG. Note that the focus driving section 31A may move the photographing focusing lens 31 along the optical axis of the photographing optical system 30 manually or by the operation of the operation section 242 by the user.

The focus driver 43A receives control from the main controller 211 and moves the OCT focus lens 43 along the optical axis of the interference optical system (the optical path of the measurement light) in the OCT unit 100 . The focus driving section 43A is provided with a holding member that holds the OCT focusing lens 43, an actuator that generates driving force for moving the holding member, and a transmission mechanism that transmits this driving force. The actuator is composed of, for example, a pulse motor. The transmission mechanism is configured by, for example, a combination of gears, a rack and pinion, or the like. As a result, the focus drive unit 43A controlled by the main control unit 211 moves the OCT focus lens 43, thereby changing the focus position of the measurement light. Note that the focus drive unit 43A may move the OCT focus lens 43 along the optical axis of the interference optical system either manually or by the user's operation on the operation unit 242 .

The main controller 211 can control the exposure time (charge accumulation time), sensitivity, frame rate, etc. of the CCD image sensor 35 . The main controller 211 can control the exposure time, sensitivity, frame rate, etc. of the CCD image sensor 38 .

The main control unit 211 can control the display of the fixation target and visual acuity measurement target on the LCD 39 . As a result, it becomes possible to switch the visual target presented to the eye E to be examined and change the type of the visual target. Further, by changing the display position of the optotype on the LCD 39, it is possible to change the optotype presenting position for the eye E to be examined.

The main control unit 211 can relatively change the difference between the optical path length of the reference light LR and the optical path length of the measurement light LS by controlling the optical path length changing unit 41 . The main control unit 211 controls the optical path length changing unit 41 so that the target portion of the eye E to be examined is rendered in a predetermined range within the frame of the OCT image. Specifically, the main control unit 211 controls the optical path length changing unit 41 so that the target portion of the eye E to be examined is drawn at a predetermined z position (position in the depth direction) within the frame of the OCT image. is possible.

The main controller 211 can change the scanning position of the measurement light LS on the fundus Ef of the eye E to be examined or the anterior segment of the eye by controlling the optical scanner 42 . The control of the optical scanner 42 includes control of the scanning position, scanning range, or scanning speed by a galvanomirror that deflects the measurement light LS in the x direction; speed control, etc. Further, the main controller 211 can control the scanning mode of the measurement light LS by the optical scanner 42 . Scanning modes of the measurement light LS include, for example, horizontal scanning, vertical scanning, cross scanning, radial scanning, circular scanning, concentric scanning, and spiral scanning.

By controlling the light source unit 101, the main control unit 211 can control switching between lighting and extinguishing of the light L0, change of the light amount of the light L0, and the like.

The main controller 211 can relatively change the difference between the optical path length of the reference light LR and the optical path length of the measurement light LS by controlling the reference driver 114A. 114 A of reference drive parts move the corner cube 114 provided in the reference optical path. Thereby, the length of the reference optical path is changed. The main control unit 211 controls the reference driving unit 114A so that the target portion of the eye E to be examined is rendered in a predetermined range within the frame of the OCT image. Specifically, the main control unit 211 can control the reference driving unit 114A so that the target region of the subject's eye E is drawn at a predetermined z position within the frame of the OCT image. The main control unit 211 can relatively change the difference between the optical path length of the reference light LR and the optical path length of the measurement light LS by controlling at least one of the optical path length changing unit 41 and the reference driving unit 114A. is. In the following description, the main control unit 211 adjusts the optical path length difference between the measurement light LS and the reference light LR by controlling only the optical path length changing unit 41. However, it is possible to control only the reference driving unit 114A. may be used to adjust the optical path length difference between the reference light LR and the measurement light LS.

The main controller 211 can control the exposure time (charge accumulation time), sensitivity, frame rate, etc. of the detector 125 . Also, the main control unit 211 can control the DAQ 130 .

An optical system driving unit (not shown) three-dimensionally moves the optical system (the optical system shown in FIGS. 1 and 2) provided in the ophthalmologic apparatus 1 . The main controller 211 can control the optical system driver so as to maintain the positional relationship between the eye E to be examined and the apparatus optical system. This control is used in alignment and tracking. Tracking is to move the apparatus optical system according to the movement of the eye E to be examined. Alignment and focusing are performed in advance when tracking is performed. Tracking is performed by moving the optical system of the device in real time according to the position and orientation of the eye to be examined E based on the image obtained by moving image shooting of the eye to be examined E, thereby maintaining a suitable positional relationship in which alignment and focus are achieved. It is a function.

(storage unit 212)
The storage unit 212 stores various data. The data stored in the storage unit 212 includes, for example, image data of an OCT image, image data of a fundus image, eye information to be examined, and processing results of the data processing unit 300 . The eye information to be examined includes subject information such as subject ID and name, left/right eye identification information, electronic medical record information, and the like.

(Image forming section 220)
The image forming unit 220 forms image data of a tomographic image of the fundus oculi Ef or the anterior segment based on the detection signal from the detector 125 sampled by the DAQ 130 . Image forming section 220 includes a processor. In this specification, "image data" and "images" based thereon may be regarded as the same.

As shown in FIG. 4 , the image forming section 220 includes a tomographic image forming section 221 .

(Tomographic image forming unit 221)
The tomographic image forming unit 221 forms image data of a tomographic image of the fundus oculi Ef based on the sampling result of the detection signal by the DAQ 130 . The tomographic image forming unit 221 can form a tomographic image based on data acquired in preliminary measurement. The processing executed by the tomogram forming unit 221 includes noise removal (noise reduction), filtering, FFT (Fast Fourier Transform), and the like, as in conventional spectral domain OCT. When another type of OCT is applied, the tomogram forming unit 221 performs known processing according to that type.

Such a tomographic image forming unit 221 can form a tomographic image based on data obtained by blood flow measurement (first scanning and/or second scanning).

For example, as disclosed in Japanese Unexamined Patent Application Publication No. 2017-79886, in the embodiment, scanning (preliminary measurement) for setting a target site for blood flow measurement and blood flow measurement for the site set from the result of the preliminary measurement are performed. Scanning for acquiring flow information (blood flow measurement) is performed.

In the preliminary measurement, multiple cross sections of the fundus oculi Ef are repeatedly scanned with the measurement light LS. The scanning of multiple cross sections is, for example, raster scanning along multiple scanning lines parallel to each other. The number of scanning lines for raster scanning is, for example, 128, thereby scanning the three-dimensional region of the fundus oculi Ef (three-dimensional scanning). Note that the scanning pattern and the number of scanning lines are not limited to this example.

In the blood flow measurement, two types of scanning (first scanning and second scanning) are performed based on the site of the fundus oculi Ef set from the result of the preliminary measurement. A part set based on the result of the preliminary measurement is called a target part. The site of interest includes a blood vessel of interest and a section of interest. The blood vessel of interest is the blood vessel of the fundus oculi Ef set from the result of preliminary measurement. A blood vessel of interest may be a blood vessel in which a pulse wave is clearly observed, such as an artery. The cross section of interest is a cross section that intersects the blood vessel of interest. Furthermore, the region of interest may include one or more cross-sections located near the cross-section of interest. The one or more cross-sections are subjected to an OCT scan to determine the inclination of the vessel of interest.

In the first scan, two or more sections that intersect the blood vessel of interest are scanned with the measurement light LS. The data acquired by the first scan are used to obtain the inclination (orientation) of the target blood vessel in the target cross section. On the other hand, in the second scan, the section of interest that intersects the blood vessel of interest is repeatedly scanned with the measurement light LS. The cross section on which the first scan is performed is placed near the cross section of interest. The second scan is Doppler measurement using OCT (Doppler OCT).

It is desirable that the target cross sections of the first scan and the second scan are oriented perpendicularly to the running direction of the blood vessel of interest in the xy plane. As shown in the fundus image D of FIG. 9, in the embodiment, for example, two cross-sections C11 and C12 on which the first scan is performed and a cross-section of interest C2 on which the second scan is performed are focused in the vicinity of the optic papilla Da. It is set to cross the blood vessel Db. One of the two cross sections C11 and C12 is positioned upstream of the target blood vessel Db with respect to the target cross section C2, and the other is positioned downstream. The distance (distance between cross sections) of each cross section C11 and C12 from the target cross section C2 is determined in advance.

The second scan is preferably performed over at least one cardiac cycle of the subject's heart. Thereby, blood flow information in all phases of the heart is obtained. The execution time of the second scan may be a preset constant time (for example, two seconds), or may be a time set for each subject or each examination. In one example of the latter, heartbeat data of the subject obtained using an electrocardiograph can be utilized.

For example, the tomographic image forming unit 221 generates a tomographic image representing the morphology of the cross section C11 and a tomographic image representing the morphology of the cross section C12 based on the detection result of the interference light LC obtained by the first scanning of the cross sections C11 and C12. to form At this time, one tomographic image can be formed by scanning the cross section C11 once, and one tomographic image can be formed by scanning the cross section C12 once. Alternatively, one tomographic image is obtained based on a plurality of tomographic images obtained by scanning the cross section C11 multiple times, and one tomographic image is obtained based on a plurality of tomographic images obtained by scanning the cross section C12 multiple times. A tomogram can be acquired. Examples of processing for obtaining one tomogram from multiple tomograms include processing for improving image quality by averaging multiple tomograms and processing for selecting an optimum one from multiple tomograms.

Further, the tomographic image forming unit 221 forms a group of tomographic images representing time-series changes in the form of the cross section of interest C2 based on the detection result of the interference light LC obtained by the second scanning of the cross section of interest C2. This processing will be described in more detail. In the second scan, the section of interest C2 is repeatedly scanned as described above. Detection signals are sequentially input to the tomogram forming unit 221 from the detector 125 of the OCT unit 100 according to the second scan. The tomographic image forming unit 221 forms one tomographic image of the cross section of interest C2 based on the detection signal group corresponding to one scan of the cross section of interest C2. The tomographic image forming unit 221 repeats this process the number of times the second scan is repeated, thereby forming a series of tomographic images in time series. Here, these tomographic images may be divided into a plurality of groups, and the tomographic images of each group may be averaged to improve image quality.

(User interface 240)
The ophthalmologic apparatus 1 is provided with a user interface 240 . User interface 240 includes display unit 241 and operation unit 242 . Display unit 241 includes display device 3 . The operation unit 242 includes various operation devices and input devices. The user interface 240 may include a device, such as a touch panel, that combines a display function and an operation function. It is also possible to construct embodiments that do not include at least a portion of user interface 240 . For example, the display device may be an external device connected to the ophthalmic equipment.

(Data processing unit 300)
The data processing unit 300 performs various data processing (image processing) and analysis processing on the image data formed by the image forming unit 220 . For example, the data processing unit 300 executes correction processing such as luminance correction and dispersion correction of an image. The data processing unit 300 also performs various image processing and analysis processing on images (anterior segment images, etc.) obtained using the CCD image sensors 35 and 38 .

The data processing unit 300 can form volume data (voxel data) of the subject's eye E by executing known image processing such as interpolation processing for interpolating pixels between tomographic images. When displaying an image based on volume data, the data processing unit 300 performs rendering processing on this volume data to form a pseudo-three-dimensional image when viewed from a specific line-of-sight direction.

The data processing unit 300 can form an image by performing various types of rendering on the acquired volume data (three-dimensional data set, stack data, etc.). Images formed by the data processing unit 300 include B-mode images (longitudinal cross-sectional images, axial cross-sectional images) at arbitrary cross-sections, C-mode images (cross-sectional images, horizontal cross-sectional images) at arbitrary cross-sections, projection images, shadowgrams, and the like. There is An arbitrary cross-sectional image, such as a B-mode image or a C-mode image, is formed by selecting pixels (pixels, voxels) on a specified cross-section from a three-dimensional data set. A projection image is formed by projecting a three-dimensional data set in a predetermined direction (z direction, depth direction, axial direction). A shadowgram is formed by projecting a portion of the three-dimensional data set (for example, partial data corresponding to a specific layer) in a predetermined direction. An image such as a C-mode image, a projection image, or a shadowgram whose viewpoint is the front side of the subject's eye is called an en-face image.

The data processing unit 300 also generates a B-mode image or a front image (vessel-enhanced image, angiographic image, etc.) in which retinal vessels and choroidal vessels are emphasized based on data (for example, B-scan image data) collected in time series by OCT. gram) can be constructed. For example, time-series OCT data can be collected by repeatedly scanning substantially the same portion of the eye E to be examined.

In some embodiments, the data processing unit 300 compares time-series B-scan images obtained by B-scans of approximately the same site, and converts the pixel values of the portions where the signal intensity changes to the pixel values corresponding to the changes. An enhanced image in which the changed portion is emphasized is constructed by the conversion. Furthermore, the data processing unit 300 extracts information for a predetermined thickness in a desired region from the constructed multiple enhanced images and constructs an en-face image to form an OCT angiogram.

Furthermore, the data processing unit 300 can specify a predetermined layer area in the tomographic image formed by the image forming unit 220 . The identifiable layer regions include the inner limiting membrane, nerve fiber layer, ganglion cell layer, inner reticular layer, inner nuclear layer, outer reticular layer, outer nuclear layer, outer limiting membrane, photoreceptor layer, retinal pigment epithelium layer, and choroid. , sclera, interfaces of each layer region, and the like.

The process of specifying a predetermined layer area from the tomogram typically includes segmentation process. Segmentation processing is a known processing for identifying partial regions in a tomogram. The data processing unit 300 performs segmentation processing, for example, based on the luminance value of each pixel in the tomographic image. That is, each layer region of the fundus oculi Ef has a characteristic reflectance, and the image regions corresponding to these layer regions also have characteristic luminance values. The data processing unit 300 can specify a target image region (layer region) by executing segmentation processing based on these characteristic luminance values. The data processing unit 300 identifies a layer region where at least a retinal artery or a retinal vein exists (for example, a layer region between the internal limiting membrane and Bruch's membrane). In some embodiments, the data processor 300 identifies image regions in the phase image (discussed below) that correspond to layer regions identified in the tomogram.

The data processing unit 300 generates blood flow information based on the image formed by the image forming unit 220, analyzes the generated blood flow information, and determines whether the subject belongs to the sick (healthy) group. It is possible to classify The blood flow information includes blood flow velocity in the fundus oculi Ef. Such a data processing unit 300 includes a phase image forming unit 310, a blood flow information generating unit 320, and a blood flow information analyzing unit 330, as shown in FIG.

(Phase image forming unit 310)
The phase image forming unit 310 forms a phase image based on data (OCT data) obtained by repeatedly performing OCT scans on a (substantially identical) predetermined region of the fundus oculi Ef. The phase image forming unit 310 forms a phase image representing time-series changes in the phase difference in the scanned cross section based on the data acquired in the preliminary measurement. In the preliminary measurement, multiple cross-sections of the fundus oculi Ef are repeatedly scanned. The phase image forming unit 310 forms a phase image in each of these cross sections based on the data obtained by preliminary measurement.

Also, the phase image forming unit 310 forms a phase image based on the data acquired by the second scan for blood flow measurement. For example, the phase image forming unit 310 forms a phase image representing the time-series change of the phase difference in the cross section of interest C2 based on the detection result of the interference light LS obtained by the second scanning of the cross section of interest C2.

The data used to form the phase image may be the same as the data used by the tomographic image forming unit 221 to form the tomographic image. In this case, it is possible to easily align the tomographic image and the phase image of the scanned cross section. That is, for a tomographic image and a phase image formed based on the same data, it is possible to naturally associate the pixels of the tomographic image with the pixels of the phase image.

An example of a method of forming a phase image will be described. The phase image of this example is obtained by calculating the phase difference between adjacent A-line complex signals (signals corresponding to adjacent scanning points). In other words, the phase image of this example is formed based on the time-series change of the pixel value (luminance value) of each pixel of the tomographic image of the scanned cross section. For any pixel, the phase imager 310 considers a graph of its luminance value over time. The phase image forming unit 310 obtains the phase difference Δφ between two points in time t1 and t2 (t2=t1+Δt) separated by a predetermined time interval Δt in this graph. Then, this phase difference Δφ is defined as the phase difference Δφ(t1) at time t1 (more generally, any time between two times t1 and t2). By executing this process for each of a large number of preset time points, a time-series change in the phase difference at the pixel is obtained.

A phase image is an image representing the phase difference value of each pixel at each point in time. This imaging processing can be realized, for example, by expressing the value of the phase difference by display color or brightness. At this time, the color (for example, red) indicating that the phase has increased along the time series can be made different from the color (for example, blue) that indicates that the phase has decreased. Also, the magnitude of the phase change amount can be expressed by the intensity of the display color. By adopting such an expression method, it is possible to present the direction and size of blood flow in colors and densities. A phase image is formed by executing the above processing for each pixel.

The time-series change of the phase difference can be obtained by sufficiently reducing the time interval Δt to secure the phase correlation. At this time, oversampling is performed by setting the time interval Δt to a value less than the time corresponding to the resolution of the tomographic image in the scanning of the measurement light LS.

(Blood flow information generator 320)
The blood flow information generation unit 320 generates a predetermined layer region (for example, , the layer region between the inner limiting membrane and Bruch's membrane).

The blood flow information generator 320 processes the data obtained by blood flow measurement of the site set based on the preliminary measurement. In this example, the blood flow information generation unit 320 obtains the inclination of the target blood vessel in the target cross section based on the data acquired by the first scan for one blood vessel. Furthermore, the blood flow information generation unit 320 obtains blood flow information regarding the target blood vessel based on the inclination of the target blood vessel obtained based on the first scan and the data obtained by the second scan. Blood flow information includes, for example, blood flow velocity and blood flow volume. As shown in FIG. 6, the blood flow information generation unit 320 includes a blood vessel region identification unit 321, an inclination calculation unit 322, a blood flow velocity calculation unit 323, a blood vessel diameter calculation unit 324, and a blood flow amount calculation unit. 325.

(Vascular region identification unit 321)
The blood vessel region identifying unit 321 identifies a blood vessel region corresponding to the target blood vessel in the tomographic image formed by the tomographic image forming unit 221 . Furthermore, the blood vessel region identifying section 321 identifies a blood vessel region corresponding to the target blood vessel in the phase image formed by the phase image forming section 310 . Identification of the blood vessel region is performed by analyzing the pixel values of each image (for example, thresholding). In some embodiments, the blood vessel region identifying unit 321 performs known image processing such as threshold processing, edge detection, binarization, thinning, and region growing on the tomogram or phase image. Identify the vascular region by executing

In some embodiments, when the shape of the blood vessel region depicted in the phase image is substantially circular, the blood vessel region identifying unit 321 identifies substantially circular regions with large changes in luminance difference or phase difference. to identify the blood vessel region.

In this embodiment, it is possible to form a tomogram and a phase image from the same data. As a result, the vascular region identifying unit 321 identifies a vascular region in the tomographic image in which the cross-sectional shape is rendered relatively clearly, and identifies a region in the phase image corresponding to the identified vascular region as the vascular region. is possible. Further, even if the tomographic image and the phase image are created from different data, if their registration is possible directly or indirectly, the identification result of the blood vessel region of one image can be used for the other image. can be used to identify the blood vessel region of the image.

In some embodiments, the blood vessel region identification unit 321 identifies the designated blood vessel region based on the operation content of the operation unit 242 by the user. The user operates the operation unit 242 while viewing the front image or the cross-sectional phase image of the fundus oculi Ef displayed on the display unit 241 by the main control unit 211 to select the blood vessel region corresponding to the desired blood vessel of interest. can be specified.

(Inclination calculator 322)
The inclination calculator 322 calculates the inclination of the target blood vessel Db in the target cross section C2 based on the data acquired by the first scan. At this time, it is also possible to further use the data obtained by the second scan. The inclination calculator 322 calculates the inclination of the target blood vessel Db in the target cross section C2 based on the cross-sectional distance and the result of specifying the blood vessel region. The inter-section distance may include the distance between the section C11 and the section C12. Also, the cross-section distance may include the distance between the cross section C11 and the cross section C2 of interest and the distance between the cross section C12 and the cross section of interest C2.

An example of a method of calculating the inclination of the target blood vessel Db will be described with reference to FIG. Tomographic images G11 and G12 are respectively a tomographic image representing a cross section C11 and a tomographic image representing a cross section C12 to which the first scanning is applied. Also, the tomographic image G2 is a tomographic image representing the target cross section C2 to which the second scanning is applied. References V11, V12, and V2 indicate the vascular region in the tomographic image G11, the vascular region in the tomographic image G12, and the vascular region in the tomographic image G2, respectively. Note that these blood vessel regions correspond to the cross section of the target blood vessel Db. In FIG. 10, the z-coordinate axis faces downward, which substantially coincides with the irradiation direction of the measurement light LS (the direction of the optical axis of the optical path of the measurement light LS, the axial direction, and the A-line direction). . Also, let L be the interval between adjacent tomographic images (cross sections).

In one example, the tilt calculator 322 calculates the tilt U of the target blood vessel Db in the target cross section C2 based on the positional relationship between the three blood vessel regions V11, V12, and V2. This positional relationship is obtained, for example, by connecting three blood vessel regions V11, V12 and V2. Specifically, the slope calculator 322 identifies feature points in each of the three blood vessel regions V11, V12, and V2, and connects these feature points. The feature points include the center position, the center of gravity position, the top (the position with the smallest z-coordinate value), the bottom (the position with the largest z-coordinate value), and the like. Further, as a method of connecting these feature points, there are a method of connecting with a line segment, a method of connecting with an approximation curve (spline curve, Bezier curve, etc.), and the like.

Further, the tilt calculator 322 calculates the tilt U based on the line connecting these feature points. When line segments are used, for example, the slope of the first line segment connecting the feature point of the blood vessel region V2 in the cross section C2 of interest and the feature point of the blood vessel region V11 in the cross section C11 and the feature point of the blood vessel region V2 The slope U is calculated based on the slope of the second line segment connecting the feature points of the blood vessel region V12 in the cross section C12. As an example of this calculation process, the average value of the slopes of two line segments can be obtained. Further, as an example of connecting with an approximated curve, the slope of the approximated curve at the intersection position of the approximated curve and the target cross-section C2 can be obtained. Note that the cross-sectional distance L is used when embedding these tomographic images G11, G12, and G2 in the xyz coordinate system in the process of obtaining line segments and approximated curves.

In this example, the vascular regions in three cross sections are considered, but it is also possible to determine the inclination by considering the vascular regions in two cross sections. As a specific example, the inclination U of the target blood vessel Db in the target cross section C2 can be obtained based on the blood vessel region V11 in the cross section C11 and the blood vessel region V12 in the cross section C12. Alternatively, it is also possible to use the slope of the first line segment or the second line segment as the slope U.

The method for calculating the inclination of the target blood vessel in the target cross section is not limited to the above. For example, the following method can be applied. First, an OCT scan is performed on a section intersecting the section of interest and along the blood vessel of interest. Next, a tomographic image is formed based on the data acquired by this OCT scanning, and the tomographic image is segmented to specify an image region (layer region) corresponding to a predetermined tissue. The specified layer region is, for example, a layer region (ILM region) corresponding to the inner limiting membrane. The inner limiting membrane is a retinal tissue that defines the boundary between the retina and the vitreous body, and is relatively clearly delineated. Furthermore, a line segment that approximates the shape of the identified layer region is obtained, and the slope thereof is obtained. The inclination of the approximated line segment is expressed, for example, as an angle with respect to the z-coordinate axis or as an angle with respect to the xy plane (that is, a plane perpendicular to the z-coordinate axis).

(Blood flow velocity calculator 323)
The blood flow velocity calculation unit 323 calculates the blood flow velocity of the blood flowing through the blood vessel Db of interest in the cross section C2 of interest based on the time series change of the phase difference obtained as the phase image. This calculation target may be the blood flow velocity at a certain point in time, or the time-series change in the blood flow velocity (blood flow velocity change information). In the former case, for example, it is possible to selectively acquire the blood flow velocity at a predetermined time phase of the electrocardiogram (for example, the time phase of the R wave). Also, the time range in the latter is the whole or an arbitrary part of the time during which the section of interest C2 is scanned.

When the blood flow velocity change information is obtained, the blood flow velocity calculation unit 323 can calculate the statistic value of the blood flow velocity in the range of time. These statistics include mean, standard deviation, variance, median, maximum, minimum, maximum and minimum. It is also possible to create a histogram of blood velocity values.

The blood flow velocity calculator 323 calculates the blood flow velocity using the Doppler OCT method as described above. At this time, the inclination U of the target blood vessel Db in the target cross section C2 calculated by the inclination calculator 322 is taken into consideration. Specifically, the slope calculator 322 uses the following equation.

In equation (1), Δf represents the Doppler shift received by the scattered light of the measurement light LS, n represents the refractive index of the medium (blood), and v represents the flow velocity (blood velocity) of the medium. θ represents the angle between the direction of incidence of the measurement light LS and the direction of flow of the medium (inclination U), and λ represents the central wavelength of the measurement light LS.

In an embodiment, n and λ are known, Δf is obtained from the time-series variation of the phase difference, and θ is obtained from the slope U (or θ is obtained as the slope U). By substituting these values into equation (1), the blood flow velocity v is calculated.

(Blood vessel diameter calculator 324)
The blood vessel diameter calculator 324 calculates the diameter of the target blood vessel Db in the target cross section C2 by analyzing the fundus image or the OCT image. When the fundus image is used, the part of the fundus oculi Ef including the position of the cross section C2 of interest is photographed, and the cross section of interest C2 is obtained based on the fundus image thus obtained (for example, a color fundus image, a red-free image, etc.). , that is, the diameter of the blood vessel region V2. When an OCT image is used, this OCT image is, for example, a tomographic image formed based on the second scan or an image formed based on preliminary measurement. Calculation of such a blood vessel diameter is performed in a conventional manner.

(Blood flow calculation unit 325)
The blood flow rate calculator 325 calculates the flow rate of blood flowing through the target blood vessel Db based on the blood flow velocity calculation result and the blood vessel diameter calculation result. An example of this processing is described below.

Assume that the blood flow in the blood vessel is Hagen-Poiseuille flow. Further, when the blood vessel diameter is w and the maximum value of the blood flow velocity is Vm, the blood flow rate Q is expressed by the following equation.

The blood flow volume calculation unit 325 substitutes the blood vessel diameter calculation result w by the blood vessel diameter calculation unit 324 and the maximum value Vm based on the blood flow velocity calculation result by the blood flow velocity calculation unit 323 into Equation (2). , the blood flow Q is calculated.

(Blood flow information analysis unit 330)
The blood flow information analysis unit 330 uses the blood flow information generated by the blood flow information generation unit 320 to perform analysis processing. In this example, the blood flow information generator 320 generates the blood flow velocity of one retinal artery or retinal vein as blood flow information. The blood flow information analysis unit 330 generates time-series data of the blood flow information (that is, data representing the time change of the blood flow velocity) representing the change over time of the blood flow information generated by the blood flow information generation unit 320. do. The blood flow information analysis unit 330 acquires the feature amount of the blood flow information by performing time-frequency analysis on the generated time-series data of the blood flow information, and determines the subject based on the acquired feature amount. Classify into one of a plurality of groups, including a healthy group and a sick group.

Such a blood flow information analysis unit 330 includes an extraction unit 340, an analysis unit 350, a classification unit 360, and a learning unit 370, as shown in FIG.

(Extraction unit 340)
The extraction unit 340 extracts blood flow information (time series data of blood flow velocity) for one beat period from the blood flow information (time series data of blood flow velocity) generated by the blood flow information generation unit 320 . The extractor 340 includes a cardiac cycle identifier 341 , an interpolator 342 and a normalizer 343 .

(Cardiac cycle identification unit 341)
The cardiac cycle identifying unit 341 identifies a cardiac cycle (one beat period) from the blood flow velocity time-series data. Blood flow velocity changes due to the heartbeat. The cardiac cycle identifying unit 341 identifies the cardiac cycle by, for example, identifying points of change in blood flow velocity corresponding to the pulsations.

FIG. 11 shows an operation explanatory diagram of the extraction unit 340. As shown in FIG. FIG. 11 shows an example of a blood flow velocity waveform, which is time-series data of blood flow velocity. In FIG. 11, the horizontal axis represents time, and the vertical axis represents blood flow velocity.

The cardiac cycle identifying unit 341 identifies the period between two timings at which the blood flow velocity reaches a predetermined value as the cardiac cycle (1 beat period) _THR required for one heartbeat from the blood flow velocity time-series data. Predetermined values include a minimum value, a maximum value, a minimum value, a maximum value, and the like. In the embodiment, the cardiac cycle identification unit 341 identifies the period between two timings at which the blood flow velocity reaches its minimum value as the cardiac cycle _THR . The cardiac cycle identification unit 341 may output the statistic values of two or more cardiac cycles identified from the blood flow velocity time-series data as the cardiac cycle _THR . These statistics include mean, standard deviation, variance, median, maximum, minimum, maximum and minimum.

(Interpolation unit 342)
The interpolation unit 342 normalizes the time-series data of the blood flow velocity during the cardiac cycle _THR specified by the cardiac cycle specifying unit 341 along the time axis. Normalization in the direction of the time axis corresponds to normalization of the cardiac cycle. Specifically, the interpolating unit 342 interpolates the time-series data of the blood flow velocity whose cardiac cycle is specified by the cardiac cycle specifying unit 341 so that the number of measurement samplings in a specified period becomes a specified number of samplings. Execute the process. The predetermined period includes, for example, one beat period, one second, two seconds, and the like. As the predetermined sampling number, there are 2 ^k (k is a natural number, eg, k=7) points. In this embodiment, the interpolation unit 342 performs interpolation processing so that the number of samplings is 2 ⁷ (=128) points (dimensions) in one beat period.

When the number of measurement samplings in the predetermined period is less than the predetermined number of samplings, the interpolator 342 can add blood flow velocities at one or more sampling points by interpolation. In some embodiments, interpolator 342 adds one or more sampling points using known interpolation methods such as linear interpolation, Lagrangian interpolation, spline interpolation, or the like.

When the number of measurement samplings in a predetermined period exceeds the predetermined number of samplings, the interpolator 342 can thin out blood flow velocities at one or more sampling points.

By normalizing the time-series data of blood flow velocity along the time axis in this way, it is possible to improve the analysis accuracy of the time-series data of multiple subjects with different heart rates (cardiac cycles). become.

(Normalization unit 343)
The normalization unit 343 normalizes the blood flow velocity time-series data to which sampling points have been added (or thinned out) by the interpolation unit 342 in the blood flow velocity axis direction. Normalization of the blood velocity axial direction corresponds to normalization of the blood velocity (amplitude). Specifically, the normalization unit 343 normalizes the blood flow velocity within the cardiac cycle _THR by the maximum value, as shown in FIG. For example, the normalization unit 343 calculates the difference _VHR between the maximum value and the minimum value of the blood flow velocity within the cardiac cycle _THR so that the maximum value is "1" and the minimum value is "0". normalize .

By normalizing the time-series data of blood flow velocity in the direction of the blood flow velocity axis in this way, the blood flow velocity of a plurality of subjects with different maximum values, minimum values, or fluctuation widths of blood flow velocity can be obtained. It is possible to improve the analysis accuracy for time-series data of

In some embodiments, the normalization processing by the normalization unit 343 is omitted for the blood velocity time-series data that has been subjected to the interpolation processing by the interpolation unit 342 . In some embodiments, the interpolator 342 is configured to perform the interpolation process described above on the time-series data of blood flow velocity normalized by the normalizer 343 .

(analysis unit 350)
The analysis unit 350 performs time-frequency analysis on the time-series data of the blood flow velocity extracted by the extraction unit 340, and obtains the feature amount of the acquired analysis data. In this embodiment, the analysis unit 350 performs a continuous wavelet transform on the time-series data of the blood flow velocity in one pulsation period, and calculates the scale value and the shift value of the mother wavelet at the feature position of the acquired wavelet transform data. Ask for Hereinafter, continuous wavelet transform is simply referred to as "wavelet transform".

Such an analysis unit 350 includes a wavelet transform unit 351, a feature position search unit 352, and a feature amount identification unit 353, as shown in FIG.

(Wavelet transform unit 351)
The wavelet transform unit 351 performs wavelet transform on the time-series data of the blood flow velocity in one pulsation period.

For example, a wavelet function Ψ _a,b (t) (t is time) used as a basis is defined as shown in Equation (3). In equation (3), "a" represents a scale value that converts the time axis to "1/a" times, and "b" represents a shift value that moves the time axis by "b".

Further, as the mother wavelet, for example, when a Mexican Hat wavelet is adopted as in the following equation (4), the wavelet function Ψ _a,b (t) in equation (3) is replaced by is represented as

Here, the time-series data of the blood flow velocity in one pulsation period extracted by the extraction unit 340 is assumed to be the analyzed signal x(t). The wavelet transform unit 351 performs wavelet transform on the signal to be analyzed x(t) according to equation (6), and converts the transformed data W _Ψ [x(t), which is a function of the scale value “a” and the shift value “b”. ].

(Characteristic position searching unit 352)
The feature position search unit 352 searches for feature positions of the transformed data W _Ψ [x(t)] obtained by the wavelet transform unit 351 . The feature position searching unit 352 searches for feature positions based on the transformed data W[psi][ _x (t)]. Examples of feature positions include the position showing the maximum of the transformed data W _Ψ [x(t)], the position showing the minimum of the transformed data W _Ψ [x(t)], and the average of the transformed data W _Ψ [x(t)]. There is a position indicating a value (median value). In this embodiment, the feature position searching unit 352 searches for the position where the transformed data W _Ψ [x(t)] is maximum as the feature position.

(Feature amount specifying unit 353)
The feature quantity specifying unit 353 specifies the feature quantity of the transformed data W _Ψ [x(t)] obtained by the wavelet transform unit 351 . Specifically, the feature quantity specifying unit 353 specifies the feature quantity of the transformed data W _Ψ [x(t)] at the feature positions searched by the feature position searching unit 352 . The features include a scale value 'a' and a shift value 'b'. In this embodiment, the feature quantity specifying unit 353 specifies the scale value “a” and the shift value “b” at the position where the transformed data W _Ψ [x(t)] is maximum as the feature quantity.

FIG. 12 shows an operation explanatory diagram of the analysis unit 350. As shown in FIG. FIG. 12 is a plot of the converted data W _Ψ [x(t)] calculated using Equation (6), with the vertical axis representing the scale value “a” and the horizontal axis representing the shift value “b”. This is a schematic representation. In addition, in FIG. 12, each of the vertical axis and the horizontal axis is standardized (normalized) by the maximum value.

That is, the wavelet transform unit 351 performs wavelet transform on the signal to be analyzed x(t) according to Equation (6). As a result, the transformed data W[psi][ _x (t)], which is a function of the scale value [a] and the shift value "b", is distributed as shown in FIG. In addition, in FIG. 12, the value of the conversion data W[psi][ _x (t)] at each coordinate position is indicated by gradation (the larger the value, the darker the value).

The feature position searching unit 352 searches for the position showing the maximum of the transformed data W _Ψ [x(t)] calculated by the wavelet transform unit 351, and specifies the feature position CP. The feature quantity specifying unit 353 specifies the scale value a0 and the shift value b0 at the feature position CP specified by the feature position searching unit 352 as feature quantities.

(Classification unit 360)
The classifying unit 360 classifies the subject into a healthy subject group and a diseased subject group based on the scale value and shift value of the mother wavelet at the feature position of the transformed data W _Ψ [x(t)] obtained by the analyzing unit 350. classified into one of several groups, including As described above, the healthy subject group is a group that has been diagnosed by a doctor as being healthy in advance, and the sick group is a group that has been previously diagnosed as being ill by a doctor.

In some embodiments, the classifier 360 classifies the subject into either the healthy group or the sick group based on the scale value and shift value.

In some embodiments, the plurality of groups includes a healthy group, a sick group, and an intermediate range of healthy and sick groups. In this case, the classification unit 360 classifies the subject into one of the healthy subject group, the diseased subject group, and the intermediate area group based on the scale value and the shift value. By classifying the subject into the intermediate range group, it is possible to prompt the subject, the examiner, the doctor, etc., to scrutinize the test results of the detailed examination of the subject particularly carefully. .

In some embodiments, the classifier 360 classifies the subject into the middle region group and the rest group based on the scale value and the shift value.

In some embodiments, the classifier 360 classifies the subject as described above by comparing at least one of the scale value and shift value described above to a predetermined threshold.

In some embodiments, the classification unit 360 uses a classification model obtained by pre-analyzing time-series data of blood flow velocities of a plurality of subjects, based on the scale value and shift value classify subjects as above. Known analysis techniques such as regression analysis (single regression analysis, multiple regression analysis), machine learning, etc. are used for prior analysis of the time-series data of blood flow velocities of multiple subjects. Analysis methods based on machine learning include learning of classification models such as linear support vector machines, nonlinear support vector machines, decision tree methods, random forest methods, and k-nearest neighbor methods. In this case, the classification unit 360 can classify the subject as described above based on the scale value and the shift value using the regression line or curve obtained by the regression analysis. Alternatively, the classifier 360 can classify the subject as described above based on the scale value and the shift value using a classification model (learned model) obtained by machine learning. In this embodiment, the classifier 360 uses a classification model that functions as a non-linear support vector machine to classify subjects as described above based on the scale and shift values described above.

(Learning unit 370)
The learning unit 370 generates a classification model that is used by the classification unit 360 to classify subjects. The learning unit 370 executes known supervised machine learning using combinations of scale values and shift values obtained from time-series data of the blood flow velocity of the subject as training data and classification results of the training data as teacher data. to generate a classification model. As for the classification results, it is determined in advance by a doctor or the like which of the plurality of groups the patient should be classified into.

The functions of the data processing unit 300 described above are realized by one or more processors. In some embodiments, the functionality of data processing unit 300 is implemented by multiple processors that perform the functionality of each unit of data processing unit 300 .

The ophthalmologic apparatus 1 or the blood flow information analysis unit 330 is an example of the "blood flow analysis apparatus" according to the embodiment. The Doppler shift Δf is an example of a “Doppler signal” according to the embodiment. The optical system from the objective lens 22 to the OCT unit 100 in FIG. 1 is an example of the "optical system" according to the embodiment.

[motion]
The operation of the ophthalmologic apparatus 1 according to the embodiment will be described.

Examples of operations of the ophthalmologic apparatus 1 are shown in FIGS. 13 to 15. FIG. FIG. 13 represents a flow diagram of an operation example of the ophthalmologic apparatus 1 . FIG. 14 shows a flow diagram of a process example of step S1 in FIG. FIG. 15 shows a flow diagram of an example of processing in step S5 of FIG. The storage unit 212 stores computer programs for realizing the processes shown in FIGS. The main control unit 211 executes the processes shown in FIGS. 13 to 15 by operating according to this computer program. It is assumed that general preparatory processes such as alignment, focus adjustment, interference sensitivity adjustment, and z-position adjustment have already been completed.

(S1: Generate blood flow information)
First, the control unit 210 (main control unit 211) causes the blood flow information generation unit 320 to generate blood flow information including the blood flow velocity of one retinal artery or retinal vein of the eye E to be examined. Subsequently, the control unit 210 controls the blood flow information analysis unit 330 to generate the blood flow velocity time series data generated by the blood flow information generation unit 320 .

Specifically, the control unit 210 controls the optical system such as the OCT unit 100 to perform OCT measurement, and causes the blood flow information generation unit 320 to generate blood flow information using the acquired measurement results. The blood flow information analysis unit 330 generates time series data of blood flow information (time series data of blood flow velocity) by arranging the blood flow information generated by the blood flow information generation unit 320 in time series. The details of step S1 will be described later.

(S2: Identify cardiac cycle)
Next, the control unit 210 causes the cardiac cycle identifying unit 341 to identify the cardiac cycle from the time-series data generated in step S1 as described above.

(S3: interpolation)
Subsequently, the control unit 210 causes the interpolating unit 342 to interpolate the time-series data of the blood flow velocity in the cardiac cycle specified in step S2 as described above.

(S4: normalization)
Subsequently, the control unit 210 causes the normalization unit 343 to normalize the blood flow velocity time series data interpolated in step S3 as described above.

(S5: analysis)
Next, in step S4, the control unit 210 causes the analysis unit 350 to perform time-frequency analysis on the time-series data of the normalized blood flow velocity. The details of step S5 will be described later.

(S6: classification)
Next, the control unit 210 classifies the subject into either the healthy subject group or the sick subject group based on the feature amount obtained by the time-frequency analysis performed in step S5. Here, the feature amount is the scale value and shift value of the mother wavelet at the feature position of the wavelet transform data.

With this, the operation of the ophthalmologic apparatus 1 is completed (end).

The process of step S1 in FIG. 13 is performed as shown in FIG.

(S11: Specify measurement position)
First, the user or control unit 210 sets the scan range for preliminary measurement. In setting the scan range, for example, an infrared observation image of the fundus oculi Ef acquired in real time or data of the fundus oculi Ef acquired in the past (a fundus image, an OCT image, an SLO image, a phase image, etc.) is referred to. be.

The control unit 210 controls the OCT unit 100 and the like to perform preliminary measurement on the set scan range (a plurality of cross sections). As an example, the set scan range is a three-dimensional area, and raster scanning of this three-dimensional area is repeatedly performed. The phase image forming unit 310 forms a phase image for each of a plurality of cross sections based on data obtained by preliminary measurement.

(S12: Set a cross section near the measurement position)
Next, the control unit 210 displays a front image of the fundus oculi Ef. The front image is, for example, an infrared observation image acquired in real time or an image acquired in the past. The user refers to the displayed information and sets a cross section (target cross section C2) to be subjected to blood flow measurement. For example, the user refers to the setting support information displayed together with the front image of the fundus oculi Ef to designate a desired arterial region (target blood vessel Db) in this front image.

In this example, the cross section of interest C2 is set manually, but the setting is not limited to this. For example, it is possible to automatically set the target blood vessel Db and target cross section C2 based on the infrared observation image of the fundus oculi Ef and the measurement site information. The target blood vessel Db is set, for example, by selecting one of the arterial regions (for example, the thickest one). Further, the target cross section C2 is set, for example, at a position where the orientation of the target blood vessel Db is within a predetermined allowable range, and orthogonal to the running direction of the target blood vessel Db.

Further, the user or the control unit 210 (or the data processing unit 300) sets cross sections C11 and C12 that are targets of the first scan.

(S13: OCT measurement)
Subsequently, the control unit 210 executes an OCT scan of the cross-sections C11 and C12 set in step S12 (first scan). Furthermore, the control unit 210 executes repetitive OCT scanning of the cross section of interest C2 set in step S12 (second scanning).

(S14: Form an image)
The control unit 210 causes the tomographic image forming unit 221 to form tomographic images G11 and G12 corresponding to the cross sections C11 and C12 based on the data acquired by the first scan in step S13. Further, the control unit 210 causes the phase image forming unit 310 to form a phase image of the cross section C2 of interest based on the data acquired by the second scanning in step S13. Further, the tomographic image forming unit 221 forms a tomographic image of the target cross section C2 based on the data.

(S15: identify the blood vessel region)
The control unit 210 specifies the target blood vessel Db in the tomographic image formed in step S14. At this time, the control unit 210 selects an image region corresponding to the target blood vessel Db in a predetermined layer region (for example, a layer region between the internal limiting membrane and Bruch's membrane) specified by the segmentation processing by the data processing unit 300. Identify as a region.

(S16: Calculate inclination of blood vessel)
Subsequently, the control unit 210 causes the inclination calculation unit 322 to calculate the inclination U of the target blood vessel Db in the target cross section C2 as described above.

(S17: Calculate blood flow velocity)
Subsequently, the control unit 210 calculates the blood flow velocity in the target cross section C2 based on the gradient U calculated based on the first scan in step S13 and the phase image acquired in the second scan in step S13. The speed calculator 323 is made to calculate.

(S18: Calculate blood vessel diameter)
Next, the control unit 210 causes the blood vessel diameter calculation unit 324 to calculate the diameter of the target blood vessel Db in the target cross section C2.

(S19: Calculate blood flow)
Subsequently, the control unit 210 causes the blood flow calculation unit 325 to calculate the flow rate of blood flowing through the target blood vessel Db based on the blood flow velocity calculated in step S17 and the blood vessel diameter calculated in step S18. .

The control unit 210 can cause the blood flow information generation unit 320 to generate blood flow information including at least one of the blood flow velocity, blood vessel diameter, and blood flow volume calculated as described above. In this embodiment, the controller 210 causes the blood flow information generator 320 to generate blood flow information including at least the blood flow velocity. Furthermore, the control unit 210 controls the blood flow information analysis unit 330 to generate the blood flow information including time-series data of the blood flow velocity over a period longer than at least one cardiac cycle generated by the blood flow information generation unit 320. generate.

With this, the processing of step S1 in FIG. 13 is completed (end).

The process of step S5 in FIG. 13 is performed as shown in FIG.

(S21: wavelet transform)
In step S5 of FIG. 13, the control unit 210 causes the wavelet transform unit 351 to perform wavelet transform on the time-series data of the blood flow velocity normalized in step S4.

As described above, the wavelet transform unit 351 performs wavelet transform using the time-series data of the blood flow velocity as the signal to be analyzed x(t) according to the equation (6), and converts the transformed data W Obtain _Ψ [x(t)].

(S22: search for maximum position)
Subsequently, the control section 210 causes the feature position searching section 352 to search for the feature positions of the transformed data W _Ψ [x(t)] obtained in step S21.

The feature position searching unit 352 searches for the position where the transformed data W _Ψ [x(t)] is maximum as a feature position (for example, the feature position CP in FIG. 12).

(S23: Identify scale value and shift value)
Next, the control unit 210 causes the feature amount specifying unit 353 to specify the scale value and the shift value (for example, the scale value a0 and the shift value b0 in FIG. 12) at the feature position specified in step S22 as the feature amount.

With this, the process of step S5 in FIG. 13 is completed (end).

Here, the time-series data of the blood flow velocity for 36 subjects preliminarily classified into the healthy group by the doctor and 8 subjects preliminarily classified into the diseased group by the doctor were obtained. An example applied to a form will be described. In addition, the subjects classified into the disease group are comprehensive chronic limb-threating ischemia (CLTI), abdominal aortic aneurysm (Abdominal Aortic Aneurysm: AAA), acute aortic dissection (Acute Aortic Dissection) , Acute Coronary Syndrome (ACS).

FIG. 16 shows an example of time-series data of blood flow velocities of subjects classified into the healthy group and time-series data of blood flow velocities of subjects classified into the sick group. In FIG. 16, each time-series data is interpolated by the interpolator 342 so that the number of samples is ²⁷ in one beat period, and normalized in the same manner as the normalization process by the normalizer 343. .

As shown in FIG. 16, it is very difficult to clearly identify the difference in blood flow velocity between the healthy subject group and the diseased subject group.

FIG. 17 shows an example of distribution of feature quantities at feature positions of wavelet transform data calculated from the time-series data of blood flow velocity shown in FIG. In FIG. 17, the horizontal axis represents the shift value and the vertical axis represents the scale value.

As shown in FIG. 17, the feature amount calculated by performing wavelet transform on the time-series data of blood flow velocity is included in either range GR0 or GR1. A range GR0 is a range in which subjects classified into the healthy subject group are distributed. A range GR1 is a range in which subjects classified into the patient group are distributed. For example, the classification unit 360 determines whether or not the feature amount calculated by wavelet transforming the time-series data of the blood flow velocity of the subject to be analyzed is included in the ranges GR0 and GR1. Classify the subject by As a result, the subject to be analyzed can be classified into either the healthy subject group or the diseased subject group with high accuracy.

As described above, the classification unit 360 uses a regression line or curve obtained by regression analysis, or a classification model (learned model) obtained by machine learning, to classify subjects to be analyzed as healthy subjects. It can be classified with high accuracy into either the group and the sick group.

A range GR2 in which the ranges GR0 and GR1 overlap is a range in which subjects in an intermediate region between healthy subjects and diseased subjects are distributed. Therefore, by determining whether or not the subject to be analyzed is included in the range GR2, the subject, It is possible to prompt the examiner or the doctor.

As described above, according to the first embodiment, the wavelet transform is performed on the time-series data of the blood flow velocity in the fundus of the eye to be inspected for one pulsation period, thereby obtaining the characteristics of the time-series data of the blood flow velocity. quantity can be obtained. By classifying the subject into one of a plurality of groups including a healthy subject group and a diseased subject group based on the acquired feature amount, the vascular state of the subject can be evaluated with high accuracy. Become.

<Second embodiment>
In the first embodiment, the case of performing wavelet transform as time-frequency analysis has been described, but the configuration according to the embodiment is not limited to this. In the second embodiment, multiresolution analysis is performed as time-frequency analysis.

The second embodiment will be described below, focusing on the differences from the first embodiment.

The configuration of the ophthalmologic apparatus according to the second embodiment differs from the configuration of the ophthalmologic apparatus 1 according to the first embodiment in that an analysis section 350a is provided instead of the analysis section 350. FIG.

(analysis unit 350a)
FIG. 18 shows an example of a functional block diagram of the analysis unit 350a.

Analysis unit 350a decomposes 2 ^k (k is a natural number of 2 or more)-dimensional blood flow velocity time-series data in one pulsation period into approximate terms and minute terms at the first depth, and m-th (m is A multiresolution analysis is performed by repeatedly decomposing the approximation term at the (m+1)th depth into the approximation term and the minute term at the (m+1)th depth up to the nth (n is a natural number, n<k) depth, and A square sum of 2 ^k−n approximate terms at n depths and a square sum of 2 ^k−(n−1) minute terms at n-th depth are obtained as feature amounts. Such an analysis unit 350a includes a multi-resolution analysis unit 351a and a feature quantity calculation unit 352a.

(Multi-resolution analysis unit 351a)
The multiresolution analysis unit 351a performs multiresolution analysis on 2 ^k -dimensional (2 ^k sampling points) blood flow velocity time-series data in one pulsation period.

Here, let f ⁽⁰⁾ (x) be the time series data of the blood flow velocity, let f ^(j) (x) (j be the depth (order)) be the approximation term that is the scaling function, and let the wavelet function Let g ^(j) (x) be a certain detail coefficient. The multi-resolution analysis unit 351a decomposes f ⁽⁰⁾ (x) according to the following equation (7) (when n=1 in equation (7)).

That is, the multiresolution analysis unit 351a decomposes f ⁽⁰⁾ (x) up to a predetermined depth (for example, depth n=5) as shown in Equation (8).

In equation (8), when the scaling function (father wavelet) and wavelet function (mother wavelet) are defined as follows, f ^(j) (x) is represented by equations (9) and (10), g ^(j) (x) is represented by equations (11) and (12).

19A and 19B are diagrams for explaining the operation of the multi-resolution analysis unit 351a. FIG. 19A schematically shows how the blood flow velocity time-series data f ⁽⁰⁾ (x) (original) is repeatedly decomposed into approximate terms and minute terms until the depth reaches “5”. Note that FIG. 19A schematically illustrates waveforms (time-series data) reproduced by approximate terms and minute terms at each depth. FIG. 19B schematically represents the number of approximate terms and the number of minute terms at each depth.

As shown in FIG. 19A, the multi-resolution analysis unit 351a divides the blood flow velocity time-series data f ⁽⁰⁾ (x) into an approximation term f ⁽¹⁾ (x) and a minute term g ⁽¹⁾ (x). (depth=1, first depth), f ⁽¹⁾ (x) is decomposed into approximate term f ⁽²⁾ (x) and minute term g ⁽²⁾ (x) (depth=2, second depth ^{) , .} After that, it is possible to decompose in the same way.

That is, the multiresolution analysis unit 351a decomposes the time-series data of the blood flow velocity in one pulsation period into an approximation term and a minute term at the first depth, and converts the approximation term at the m-th depth to the (m+1)-th depth. A multi-resolution analysis is performed by repeating decomposition into approximation terms and minute terms up to the n-th depth (predetermined depth).

As shown in FIG. 19B, at the first depth, the 2 ^k -dimensional original time series data (original) having 2 ^k (for example, k=7) terms are combined with 2 ^k−1 dimensional approximation terms and 2 k−1 approximation terms. is decomposed into ^k -dimensional fractional terms. At the second depth, the 2 ^k−1 dimensional approximation terms at the first depth are decomposed into 2 ^k−2 dimensional approximation terms and 2 ^k−1 dimensional minute terms. At the nth depth, the 2 ^k−(n−1) dimensional approximation term at the (n−1)th depth is decomposed into 2 ^k−n dimensional approximation terms and 2 ^k−(n−1) dimensional minute terms. be done. That is, the number of dimensions (number of terms) decreases as the depth increases. The relationship between the number of approximate terms and the number of minute terms is derived from equations (7) and (8).

Note that it is possible to determine the number of repetitions (order, depth) of the decomposition of the approximation term and the minute term based on the reproduced waveform at each depth shown in FIG. 19A. Alternatively, it is possible to determine the number of repetitions of the decomposition of the approximation term and the minute term based on the difference between the reproduced waveform at each depth and the reproduced waveform at the immediately preceding depth.

(Feature amount calculator 352a)
The feature quantity calculation unit 352a calculates a feature quantity based on the approximation term and the minute term resolved to a predetermined depth by the multiresolution analysis unit 351a.

The feature amount calculation unit 352a calculates the sum of squares of 2 ^k−n (dimension) terms of the approximation terms at the n-th depth obtained by the multi-resolution analysis unit 351a, as shown in equations (13) and (14). A _n and the sum of squares D _n of 2 ^k−(n−1) (dimensions) of minute terms at the n-th depth are calculated as feature amounts. In Equation (13), a _q ⁿ represents the q-th term of the approximation terms at the n-th depth. In equation (14), d _q ⁿ represents the qth term of the microterms at the nth depth. For example, n=5 and k=7.

The operation of the ophthalmologic apparatus according to the second embodiment is substantially the same as the operation of the ophthalmologic apparatus 1 according to the first embodiment. The process of step S5 in FIG. 13 is performed as shown in FIG.

FIG. 20 shows a flow chart of a processing example of step S5 in FIG. The storage unit 212 stores computer programs for realizing the processes shown in FIGS. 13, 14, and 20. FIG. The main control unit 211 executes the processes shown in FIGS. 13, 14 and 20 by operating according to this computer program.

(S31: Multi-resolution analysis)
In step S5 of FIG. 13, the control unit 210 causes the multiresolution analysis unit 351a to perform multiresolution analysis on the time-series data of the blood flow velocity normalized in step S4.

As described above, the multi-resolution analysis unit 351a performs multi-resolution analysis on the blood flow velocity time-series data according to equations (7) to (12). As above, for example n=5 and k=7.

(S32: Calculate sum of squares of approximate terms and sum of squares of minute terms)
Subsequently, the control unit 210 causes the feature amount calculation unit 352a to calculate the feature amount using the approximation term and the minute term obtained in step S31.

As described above, the feature amount calculation unit 352a calculates the sum of squares of 2 ⁷⁻⁵ (=2 ² ) approximate terms at the fifth depth obtained in step S31 and 2 ^{7 −4} (=2 ³ ) terms and the sum of squares are calculated as feature quantities.

With this, the process of step S5 in FIG. 13 is completed (end).

Here, similarly to the first embodiment, the blood flow was measured for 36 subjects preliminarily classified into the healthy subject group and 8 subjects preliminarily classified into the diseased subject group (see FIG. 16). An example in which speed time series data is applied to the second embodiment will be described.

^FIG . 21 shows the feature values calculated by performing multiresolution analysis on the time-series data of blood flow velocity shown in FIG. An example of the distribution of the sum of squares of ²³ terms of . In FIG. 21, the horizontal axis represents the ^square sum of 22 ^approximate terms at the fifth depth, and the vertical axis represents the square sum of 23 minute terms at the fifth depth.

As shown in FIG. 21, the feature amount calculated by performing multi-resolution analysis on the time-series data of blood flow velocity is included in either range GR10 or GR11. A range GR10 is a range in which subjects classified into the healthy subject group are distributed. A range GR11 is a range in which subjects classified into the patient group are distributed. For example, the classification unit 360 determines whether or not the feature amount calculated by performing multiresolution analysis on the time-series data of the blood flow velocity of the subject to be analyzed is included in the ranges GR10 and GR11. Classify the subject by As a result, the subject to be analyzed can be classified into either the healthy subject group or the diseased subject group with high accuracy.

As described above, the classification unit 360 uses a regression line or curve obtained by regression analysis, or a classification model (learned model) obtained by machine learning, to classify subjects to be analyzed as healthy subjects. It can be classified with high accuracy into either the group and the sick group.

In the second embodiment, the learning unit 370 obtains the sum of squares of the approximation terms at the fifth depth obtained from the time ^- series data of the blood flow velocity of the subject and the sum of squares at the fifth depth The function of the classification unit 360 is performed by executing known supervised machine learning using a combination of sums of squares of 2 ^k-(n-1) terms of the minute terms as training data and using the classification results of the training data as teacher data. Generate a classification model that implements Here, the classification results are obtained by determining which of the plurality of groups the patient should be classified into in advance by a doctor or the like.

A range GR12 in which the ranges GR10 and GR11 overlap is a range in which subjects in an intermediate region between healthy subjects and diseased subjects are distributed. Therefore, by determining whether or not the subject to be analyzed is included in the range GR12, the subject, It is possible to prompt the examiner or the doctor.

As described above, according to the second embodiment, multi-resolution analysis is performed on the time-series data of the blood flow velocity for one pulsation period in the fundus of the eye to be examined. Features can be obtained. By classifying the subject into one of a plurality of groups including a healthy subject group and a diseased subject group based on the acquired feature amount, the vascular state of the subject can be evaluated with high accuracy. Become.

<Third Embodiment>
In the above embodiment, a case where wavelet transform or multi-resolution analysis is performed as time-frequency analysis has been described, but the configuration according to the embodiment is not limited to this. The ophthalmologic apparatus according to the third embodiment is configured to combine wavelet transform and multi-resolution analysis as time-frequency analysis to classify subjects into either a healthy subject group or a diseased subject group.

The third embodiment will be described below, focusing on the differences from the first embodiment or the second embodiment.

The configuration of the ophthalmologic apparatus according to the third embodiment differs from that of the ophthalmologic apparatus 1 according to the first embodiment or the ophthalmologic apparatus according to the second embodiment in that an analysis unit 350b is provided instead of the analysis unit 350 or the analysis unit 350a. The point is that

(analysis unit 350b)
FIG. 22 shows an example of a functional block diagram of the analysis unit 350b. In FIG. 22, parts similar to those in FIG. 8 or FIG.

Analysis unit 350b includes a first analysis unit 361b and a second analysis unit 362b. As in the first embodiment, the first analysis unit 361b acquires the feature amount of the time-series data of blood flow velocity by performing wavelet transform on the time-series data of blood flow velocity. As in the second embodiment, the second analysis unit 362b acquires the feature amount of the time-series data of blood flow velocity by performing multi-resolution analysis on the time-series data of blood flow velocity. That is, the first analysis unit 361b includes a wavelet transform unit 351, a feature position search unit 352, and a feature amount identification unit 353. The second analysis unit 362b includes a multi-resolution analysis unit 351a and a feature quantity calculation unit 352a.

Specifically, the first analysis unit 361b performs wavelet transform on the time-series data of the blood flow velocity in one pulsation period of the subject, and scales the mother wavelet at the feature position of the acquired wavelet transform data. A value and a shift value are obtained as feature amounts. The second analysis unit 362b performs multiresolution analysis on the time-series data of blood flow velocity, which is the analysis target of the first analysis unit 361b, and calculates the sum of squares of 2 ^k−n approximate terms at the n-th depth. and the sum of squares of 2 ^k−(n−1) minute terms at the n-th depth as feature quantities.

In the third embodiment, the classification unit 360 uses the scale value and shift value of the mother wavelet at the feature position of the transformed data W _Ψ [x(t)] obtained by the first analysis unit 361b, and the Based on the obtained sum of squares of 2 ^k−n terms of the approximation term at the n-th depth and the sum of squares of 2 ^k-(n-1) terms of the minute term at the n-th depth, the subject are classified into one of several groups, including a healthy group and a sick group.

In some embodiments, the classification unit 360 determines that the subject is included in the range GR0 based on the feature amount obtained by the first analysis unit 361b and that the feature amount obtained by the second analysis unit 362b When the subject is included in the range GR10 based on the above, the subject is classified into the healthy subject group. Further, the classification unit 360 determines that the subject is included in the range GR1 based on the feature amount obtained by the first analysis unit 361b, and determines that the subject is included in the range GR1 based on the feature amount obtained by the second analysis unit 362b. is included in the range GR11, the subject is classified into the patient group.

Further, the classification unit 360 determines that the subject is included in the range GR1 based on the feature amount obtained by the first analysis unit 361b, and determines that the subject is included in the range GR1 based on the feature amount obtained by the second analysis unit 362b. is included in the range GR10, or the subject is included in the range GR0 based on the feature amount obtained by the first analysis unit 361b, and the subject based on the feature amount obtained by the second analysis unit 362b When the examinee is included in the range GR11, the examinee is classified into the middle region group.

The learning unit 370 obtains the scale value and shift value obtained from the time-series data of the blood flow velocity of the subject, the square sum of 2 ^k−n approximate terms at the n-th depth, and the minute term at the n-th depth. The function of the classification unit 360 is performed by executing known supervised machine learning using a combination of sums of squares of 2 ^k-(n-1) terms as training data and using the classification result of the training data as teacher data. It is possible to generate a classification model that implements. Here, the classification results are obtained by determining which of the plurality of groups the patient should be classified into in advance by a doctor or the like.

As described above, according to the third embodiment, by performing wavelet transform and multi-resolution analysis on the time-series data of the blood flow velocity in one pulsation period in the fundus of the eye to be examined, the time series of the blood flow velocity Get multiple features of series data. By combining a plurality of acquired feature values and classifying the subject into one of a plurality of groups including a healthy subject group and a diseased subject group, the vascular state of the subject can be evaluated with high accuracy. become able to.

<Fourth Embodiment>
In the first to third embodiments, cases where time-frequency analysis is performed on time-series data of blood flow velocity have been described, but the configuration according to the embodiments is not limited to this. The ophthalmologic apparatus according to the fourth embodiment is configured to classify subjects into either a healthy subject group or a diseased subject group by performing regression analysis on time-series data of blood flow velocities.

The fourth embodiment will be described below, focusing on differences from the first embodiment.

The configuration of the ophthalmologic apparatus according to the fourth embodiment differs from that of the ophthalmologic apparatus 1 according to the first embodiment in that a blood flow information analysis section 330c is provided instead of the blood flow information analysis section 330 .

(Blood flow information analysis unit 330c)
FIG. 23 shows an example of a functional block diagram of the blood flow information analysis section 330c. In FIG. 23, the same parts as in FIG. 7 are denoted by the same reference numerals, and description thereof will be omitted as appropriate.

The blood flow information analysis unit 330c includes an extraction unit 340, a standardization unit 380c, a classification unit 360c, and a learning unit 370c. The extractor 340 includes a cardiac cycle identifier 341 , an interpolator 342 and a normalizer 343 . The standardization unit 380c standardizes the blood flow velocity time series data extracted by the extraction unit 340 . The classification unit 360c classifies the subject into one of a plurality of groups including a healthy subject group and a diseased subject group based on the blood flow velocity time series data standardized by the standardization unit 380c. The learning unit 370c generates a classification model that implements the function of the classification unit 360c.

(Standardization unit 380c)
The standardization unit 380c converts the distribution of the blood flow velocity time series data into a standard normal distribution. Let x be each data constituting the time-series data, μ be the average value of the time-series data, and σ be the standard deviation of the time-series data. Transform to a standard normal distribution with standardized variable z transformed according to (15).

(Classification unit 360c)
The classification unit 360c classifies the subject into one of a plurality of groups including the healthy subject group and the diseased subject group based on the time-series data of the blood flow velocity converted into the standard normal distribution by the standardization unit 380c.

In some embodiments, the plurality of groups includes a healthy group, a sick group, and an intermediate range of healthy and sick groups. In this case, the classification unit 360c classifies the subject into one of the healthy subject group, the diseased subject group, and the intermediate area group based on the scale value and the shift value.

In some embodiments, the classification unit 360c classifies the subject into either the healthy subject group or the diseased subject group based on the time-series data of the blood flow velocity converted to the standard normal distribution by the standardization unit 380c. do.

In some embodiments, the classifying unit 360c classifies the subject into the intermediate region group and other groups based on the time series data of the blood flow velocity converted to the standard normal distribution by the standardizing unit 380c. .

In some embodiments, the classification unit 360c classifies the subjects as described above using a classification model obtained by pre-analyzing time-series data of blood flow velocities of a plurality of subjects. do. For pre-analysis of time-series data of blood flow velocities of multiple subjects, known analyzes such as linear support vector machines, nonlinear support vector machines, decision tree methods, random forest methods, k-neighborhood methods, and other machine learning methods method is used. In this case, the classification unit 360c can classify the subject as described above using a classification model (learned model) obtained by machine learning.

(Learning unit 370c)
The learning unit 370c generates a classification model that is used for classification of subjects by the classification unit 360c. The learning unit 370c generates a classification model by executing known supervised machine learning using the time-series data of blood flow velocity of the subject as training data and the classification result of the training data as training data. As for the classification results, it is determined in advance by a doctor or the like which of the plurality of groups the patient should be classified into.

The functions of the blood flow information analysis unit 330c described above are realized by one or more processors. In some embodiments, the functions of the blood flow information analysis section 330c are realized by multiple processors executing the functions of each section.

As described above, according to the fourth embodiment, subjects are divided into a healthy subject group and a diseased subject group by machine learning with respect to the time-series data of the blood flow velocity in one pulsation period in the fundus of the eye to be examined. classified into one of several groups, including In particular, even if the effect of machine learning on the classification unit 360c is reduced due to the difference in absolute value distribution between explanatory variables (inputs) that are different from each other, the distribution of absolute values between explanatory variables should be uniform. Therefore, the accuracy of machine learning for the classifier 360c can be improved.

<Modification>
In the above embodiment, the mother wavelet represented by Equation (4) has been described as an example, but the configuration according to the embodiment is not limited to this. For example, in the above embodiment, any mother wavelet of equations (16) to (22) may be used.

Equation (16) represents an example of a Mexican hat mother wavelet (mexh) different from Equation (4).

Equation (17) represents an example of the Morlet mother wavelet (morl).

Equation (18) represents an example of a Complex Morlet mother wavelet (cmorBC). In equation (18), B represents the bandwidth and C represents the center frequency.

Equation (19) represents an example of a Gaussian Derivative mother wavelet (gausP). In formula (19), C represents a constant.

Equation (20) represents an example of a complex Gaussian differential mother wavelet (cgausP). In formula (20), C represents a constant.

Equation (21) represents an example of a Shannon wavelet (shanB-C). In equation (21), B represents the bandwidth and C represents the center frequency.

Equation (22) represents an example of a frequency B-spline wavelet (fbsp). (22), M represents the order of the spline, B represents the bandwidth, and C represents the center frequency.

The ophthalmologic apparatus according to the third embodiment is configured to classify subjects into the plurality of groups based on two or more feature amounts obtained by wavelet transform to which two or more mother wavelets different from each other are applied. may have been Alternatively, based on two or more feature amounts obtained by wavelet transform to which two or more mother wavelets different from each other are applied and the feature amount obtained by multiresolution analysis, subjects are divided into the above plurality of groups. It may be configured to classify.

[Action]
A blood flow analysis device, an ophthalmologic device, a blood flow analysis method, and a program according to embodiments will be described.

A blood flow analysis device (blood flow information analysis unit 330, ophthalmologic device 1) according to some embodiments includes analysis units (350, 350a) and a classification unit (360). The analysis unit performs time-frequency analysis on the time-series data of the blood flow velocity in the fundus (Ef) of the subject. The classification unit classifies the subject into one of a plurality of groups including a healthy subject group and a diseased subject group based on the feature amount of the analysis data obtained by the analysis unit.

According to this aspect, the feature amount is obtained by performing time-frequency analysis on the time-series data of the blood flow velocity, and the subject is classified based on the obtained feature amount. It is possible to accurately and highly accurately evaluate the examiner's blood vessel condition.

In some embodiments, the analysis unit performs a wavelet transform on the time-series data in one beat period, and the classifier performs a mother wavelet at feature positions (CP) of the wavelet transform data obtained by the analysis unit. A subject is classified into one of a plurality of groups based on the scale value and the shift value.

According to this aspect, since the scale value and the shift value of the mother wavelet at the feature position of the wavelet transform data obtained by the wavelet transform are obtained as feature amounts, it is useful for highly accurate evaluation of the blood vessel state. It becomes possible to easily acquire the feature amount.

In some embodiments, the analysis unit decomposes the time-series data in one beat period into an approximate term and a minute term at the first depth, and separates the approximate term at the m-th (m is a natural number of 2 or more) depth. Multi-resolution analysis is performed by repeating decomposition into approximation terms and minute terms at the (m+1)-th depth up to the n-th (n is a natural number) depth, and the classification unit determines the approximate terms at the n-th depth obtained by the analysis unit. The subject is classified into one of a plurality of groups based on the sum of squares and the sum of squares of the minute terms at the n-th depth.

According to this aspect, the sum of squares of the approximation terms at the n-th depth obtained by the multiresolution analysis (for example, the sum of squares of 2 ^k−n terms of the approximation terms, where n<k, k is 2 or more natural number) and the sum of squares of the minute terms at the n-th depth (for example, the sum of squares of 2 ^k-(n−1) terms of the minute terms) are obtained as feature quantities. It is possible to easily acquire feature quantities useful for evaluation.

In some embodiments, the analysis unit includes a first analysis unit (361b) that performs wavelet transform on the time-series data in one beat period, and an approximation term at the first depth for the time-series data in one beat period. and minute terms, and the approximation term at the mth (m is a natural number of 2 or more) depth is decomposed into the approximation term and the minute term at the (m+1)th depth up to the nth (n is a natural number) depth a second analysis unit (362b) that performs iterative multi-resolution analysis. The classification unit calculates the scale value and shift value of the mother wavelet at the feature position of the wavelet transform data obtained by the first analysis unit, and the sum of squares of the approximation terms at the nth depth and the nth depth The subject is classified into one of a plurality of groups based on the sum of squares of the fractional terms in .

According to this aspect, the scale value and the shift value of the mother wavelet at the feature position of the wavelet transform data obtained by the wavelet transform, and the sum of squares of the approximation term at the n-th depth obtained by the multiresolution analysis (for example, sum of squares of 2 ^k−n terms of approximation terms, n<k, k is a natural number of 2 or more) and sum of squares of minute terms at n-th depth (for example, 2 ^k−(n−1) of minute terms The sum of squares of the terms of ) is obtained as a feature quantity, so it is possible to easily obtain a feature quantity useful for highly accurate evaluation of the vascular condition.

In some embodiments, the feature location is the location where the wavelet transform data is the maximum.

According to such an aspect, it becomes possible to easily search for the feature position and obtain the feature amount more simply.

In some embodiments, blood flow velocity is obtained based on Doppler signals due to blood flow obtained by performing Doppler OCT on the fundus.

According to this aspect, it is possible to highly accurately evaluate the blood vessel condition of the subject's eye using an existing ophthalmologic apparatus capable of measuring blood flow velocity by OCT.

Some embodiments include an optical system (optical system from the objective lens 22 to the OCT unit 100 in FIG. 1) that performs OCT on the measurement site on the fundus at different timings, and a plurality of signals obtained by the optical system and the blood flow analysis device described above.

According to such an aspect, it is possible to provide an ophthalmologic apparatus capable of correctly and highly accurately evaluating the vascular condition of a subject.

Some embodiments include an analysis step of performing time-frequency analysis on time-series data of blood flow velocity in the fundus of the subject (Ef), and based on the feature amount of the analysis data obtained in the analysis step, and a classification step of classifying the subject into one of a plurality of groups including a healthy subject group and a diseased subject group.

According to this aspect, the feature amount is obtained by performing time-frequency analysis on the time-series data of the blood flow velocity, and the subject is classified based on the obtained feature amount. It is possible to accurately and highly accurately evaluate the examiner's blood vessel condition.

In some embodiments, the analyzing step performs wavelet transform on the time-series data in one beat period, and the classifying step performs the scale value of the mother wavelet at the feature position of the wavelet transform data obtained in the analyzing step. Classify the subject into one of a plurality of groups based on the shift value.

According to this aspect, since the scale value and the shift value of the mother wavelet at the feature position of the wavelet transform data obtained by the wavelet transform are obtained as feature amounts, it is useful for highly accurate evaluation of the blood vessel state. It becomes possible to easily acquire the feature amount.

In some embodiments, the analysis step decomposes the time-series data in one beat period into an approximate term and a minute term at the first depth, and an approximate term at the m-th (m is a natural number of 2 or more) depth. Multi-resolution analysis is performed by repeating decomposition into approximation terms and minute terms at the (m+1)-th depth up to the n-th (n is a natural number) depth, and the classification step is performed on the approximation term at the n-th depth obtained in the analysis step. The subject is classified into one of a plurality of groups based on the sum of squares and the sum of squares of the minute terms at the n-th depth.

According to this aspect, the sum of squares of the approximation terms at the n-th depth obtained by the multiresolution analysis (for example, the sum of squares of 2 ^k−n terms of the approximation terms, where n<k, k is 2 or more natural number) and the sum of squares of the minute terms at the n-th depth (for example, the sum of squares of 2 ^k-(n−1) terms of the minute terms) are obtained as feature quantities. It is possible to easily acquire feature quantities useful for evaluation.

In some embodiments, the analysis step includes a first analysis step of performing a wavelet transform on the time-series data in one beat period; and repeats the decomposition of the approximation term at the mth (m is a natural number of 2 or more) depth into the approximation term and the minute term at the (m+1)th depth up to the nth (n is a natural number) depth and a second analysis step of performing analysis. In the classification step, the scale value and the shift value of the mother wavelet at the feature position of the wavelet transform data obtained in the first analysis step, and the sum of squares of the approximation terms at the n-th depth obtained in the second analysis step and the n-th depth The subject is classified into one of a plurality of groups based on the sum of squares of the fractional terms in .

According to this aspect, the scale value and the shift value of the mother wavelet at the feature position of the wavelet transform data obtained by the wavelet transform, and the sum of squares of the approximation term at the n-th depth obtained by the multiresolution analysis (for example, sum of squares of 2 ^k−n terms of approximation terms, n<k, k is a natural number of 2 or more) and sum of squares of minute terms at n-th depth (for example, 2 ^k−(n−1) of minute terms The sum of squares of the terms of ) is obtained as a feature quantity, so it is possible to easily obtain a feature quantity useful for highly accurate evaluation of the vascular condition.

In some embodiments, the feature location is the location where the wavelet transform data is the maximum.

According to such an aspect, it becomes possible to easily search for the feature position and obtain the feature amount more simply.

In some embodiments, blood flow velocity is obtained based on Doppler signals due to blood flow obtained by performing Doppler OCT on the fundus.

According to this aspect, it is possible to highly accurately evaluate the blood vessel condition of the subject's eye using an existing ophthalmologic apparatus capable of measuring blood flow velocity by OCT.

Some embodiments are programs that cause a computer to execute each step of the blood flow analysis method described above.

According to this aspect, the feature amount is obtained by performing time-frequency analysis on the time-series data of the blood flow velocity, and the subject is classified based on the obtained feature amount. It becomes possible to provide a program capable of correctly evaluating the vascular condition of an examiner with high accuracy.

It is possible to create a computer-readable non-transitory recording medium recording the program according to any of the embodiments. This non-transitory recording medium may be in any form, examples of which include magnetic disks, optical disks, magneto-optical disks, and semiconductor memories.

The embodiment described above is merely an example of the present invention. A person who intends to implement the present invention can arbitrarily make modifications (omission, substitution, addition, etc.) within the scope of the gist of the present invention.

1 ophthalmic apparatus 100 OCT unit 210 control unit 211 main control unit 300 data processing unit 310 phase image forming unit 320 blood flow information generation units 330, 330c blood flow information analysis unit 340 extraction units 350, 350a, 350b analysis unit 351 wavelet transformation unit 351a multi-resolution analysis unit 352 feature position search unit 352a feature amount calculation unit 353 feature amount identification units 360 and 360c classification unit 361b first analysis unit 362b second analysis units 370 and 370c learning unit 380c standardization unit E subject eye Ef fundus

Claims (14)

an analysis unit that performs time-frequency analysis on time-series data of blood flow velocity in the fundus of the subject;
a classification unit that classifies the subject into one of a plurality of groups including a healthy subject group and a diseased subject group based on the feature amount of the analysis data obtained by the analysis unit;
A blood flow analyzer, comprising: The analysis unit performs wavelet transform on the time-series data in one beat period,
The classification unit classifies the subject into one of the plurality of groups based on the scale value and shift value of the mother wavelet at the feature position of the wavelet transform data obtained by the analysis unit. The blood flow analysis device according to claim 1. The analysis unit decomposes the time-series data in one beat period into an approximate term and a minute term at a first depth, and divides the approximate term at an m-th (m is a natural number of 2 or more) depth into an (m+1)-th depth. Perform multiresolution analysis that repeats decomposition into approximate terms and minute terms in n-th (n is a natural number) depth,
The classification unit classifies the subject into one of the plurality of groups based on the sum of squares of approximate terms at the n-th depth and the sum of squares of minute terms at the n-th depth obtained by the analysis unit. The blood flow analysis device according to claim 1, characterized in that it is classified into: The analysis unit is
a first analysis unit that performs a wavelet transform on the time-series data in one beat period;
The time-series data in one beat period is decomposed into an approximate term and a minute term at the first depth, and the approximate term at the m-th (m is a natural number of 2 or more) depth is divided into the approximate term and the minute term at the (m+1)-th depth. a second analysis unit that performs multi-resolution analysis that repeats decomposition into terms up to the n-th (n is a natural number) depth;
including
The classifying unit is a scale value and a shift value of the mother wavelet at the feature position of the wavelet transform data obtained by the first analyzing unit, and the sum of squares of the approximation term at the n-th depth obtained by the second analyzing unit and the sum of squares of the minute terms at the n-th depth, the subject is classified into one of the plurality of groups. The blood flow analysis apparatus according to claim 2 or 4, wherein the characteristic position is a position where the wavelet transform data is maximum. The blood flow velocity according to any one of claims 1 to 5, wherein the blood flow velocity is obtained based on a Doppler signal by blood flow obtained by performing Doppler OCT on the fundus. blood flow analyzer. an optical system that performs OCT at different timings on a measurement site on the fundus;
a blood flow information generator that obtains a blood flow velocity in the fundus based on a plurality of signals obtained by the optical system;
a blood flow analysis device according to claim 6;
An ophthalmic device, comprising: an analysis step of performing time-frequency analysis on the time-series data of the blood flow velocity in the fundus of the subject;
A classification step of classifying the subject into one of a plurality of groups including a healthy subject group and a diseased subject group based on the feature amount of the analysis data obtained in the analysis step;
A blood flow analysis method, comprising: The analyzing step performs wavelet transform on the time-series data in one beat period,
The classification step classifies the subject into one of the plurality of groups based on the scale value and shift value of the mother wavelet at the feature position of the wavelet transform data obtained in the analysis step. The blood flow analysis method according to claim 8. The analysis step decomposes the time-series data in one beat period into an approximate term and a minute term at a first depth, and converts the approximate term at an m-th (m is a natural number of 2 or more) depth to the (m+1)-th depth. Perform multiresolution analysis that repeats decomposition into approximate terms and minute terms in n-th (n is a natural number) depth,
The classification step classifies the subject into one of the plurality of groups based on the sum of squares of approximate terms at the n-th depth and the sum of squares of minute terms at the n-th depth obtained in the analysis step. 9. The blood flow analysis method according to claim 8, wherein the classification is as follows. The analysis step includes:
a first analysis step of performing a wavelet transform on the time-series data in one beat period;
The time-series data in one beat period is decomposed into an approximate term and a minute term at the first depth, and the approximate term at the m-th (m is a natural number of 2 or more) depth is divided into the approximate term and the minute term at the (m+1)-th depth. a second analysis step of performing multi-resolution analysis in which the decomposition into terms is repeated up to the n-th (n is a natural number) depth;
including
In the classification step, the scale value and the shift value of the mother wavelet at the feature position of the wavelet transform data obtained in the first analysis step, and the sum of squares of the approximation term at the n-th depth obtained in the second analysis step and a sum of squares of minute terms at the nth depth, the subject is classified into one of the plurality of groups. The blood flow analysis method according to claim 9 or 11, wherein the feature position is a position where the wavelet transform data is maximum. The blood flow velocity according to any one of claims 8 to 12, wherein the blood flow velocity is obtained based on a Doppler signal by blood flow obtained by performing Doppler OCT on the fundus. blood flow analysis method. A program for causing a computer to execute each step of the blood flow analysis method according to any one of claims 8 to 13.

Images