OCT image processing apparatus and OCT image processing method

Abstract

This technology enables the removal of speckles from OCT images using a small number of OCT scans. [Solution] The OCT image processing method (S1) includes an SCCP process (S11) in which at least one processor repeatedly calculates the shift conjugate product, which is the product of complex interference image data of a measurement target obtained by irradiating the measurement target with coherent incident light and the complex conjugate image of shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount, multiple times with varying shift amounts; an averaging process (S12) which averages the real part data of multiple shift conjugate products; and an imaging process (S13) which converts the averaged real part data into an image and outputs it.

Description

Technical Field

[0001] The present invention relates to an OCT image processing apparatus and an OCT image processing method.

Background Art

[0002] OCT (Optical Coherence Tomography or Optical Coherence Tomograph) is a biomedical tomography imaging technology using optical interference, and is widely used in ophthalmic clinical diagnosis and the field of cardiovascular clinical medicine. In recent years, it has also attracted attention as a three-dimensional microscopy technology in the fields of basic medicine, biology, and pharmacy. Although OCT is widely clinically applied in this way, a pattern of high-intensity granular signals called "speckle" appears in OCT images. Therefore, the image contrast of the actual measurement target (mainly biological tissue) is low. In addition, the size of the granular pattern of speckle is on the same order as the optically defined resolution (nominal resolution) of OCT. Therefore, in actual OCT images, it is not possible to measure fine structures on the order of the nominal resolution of the measured tissue, and the actual resolution deteriorates to about 10 times the nominal resolution. That is, only structures having a size of about 10 times the nominal resolution can be recognized.

[0003] In order to solve such problems, for example, methods such as averaging of a plurality of images (for example, Non-Patent Document 1), methods using deep learning, etc. have been studied.

Prior Art Documents

Non-Patent Documents

[0004]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] For example, image averaging techniques involve acquiring multiple images (e.g., several dozen to several hundred) from the same part of the sample being measured and statistically removing speckles by averaging them. This technique is widely used because of its good speckle removal performance. However, this method has the following two problems: (1) Acquiring the multiple images required by this method takes several tens to several hundreds of times longer than normal OCT imaging. Therefore, this method can be used for acquiring two-dimensional tomographic images but cannot be used for acquiring three-dimensional tomographic images. (2) Before averaging a large number of images, it is necessary to detect and correct the misalignment between all the images.

[0006] Of the above problems, (1) is particularly critical. For example, 3D OCT imaging without speckle removal can be completed in about 2 seconds. However, acquiring a 3D OCT image with speckle removed using the image averaging method described above requires about 1 minute of imaging time. This is a lengthy measurement time that is practically impractical for clinical testing. Furthermore, in the measurement of cultured tissues in the fields of basic medicine, biology, and pharmacy, such long measurement times are unacceptable because it is necessary to repeatedly measure many samples.

[0007] One aspect of the present invention aims to realize a technique for removing speckles appearing in OCT images using fewer OCT images. [Means for solving the problem]

[0008] To solve the above problems, an OCT image processing method according to one aspect of the present invention includes an SCCP process in which at least one processor repeatedly calculates the shift conjugate product, which is the product of complex interference image data of a measurement target obtained by irradiating the measurement target with coherent incident light and the complex conjugate image of shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount, by changing the shift amount multiple times; an averaging process that averages the real part data of the multiple shift conjugate products; and an imaging process that converts the averaged real part data into an image and outputs it.

[0009] To solve the above problems, an OCT image processing apparatus according to one aspect of the present invention includes: an acquisition unit that acquires complex interference image data of a measurement target obtained by irradiating the measurement target with coherent incident light; an SCCP processing unit that repeats the process of deriving a shift conjugate product, which is the product of the complex interference image data and a complex conjugate image of shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount, multiple times with different shift amounts; an averaging unit that averages the real part data of a plurality of the shift conjugate products; and an imaging unit that converts the averaged real part data into an image and outputs it.

[0010] Each aspect of the present invention may be implemented by a computer, in which case a control program for the OCT image processing device that enables the computer to implement the OCT image processing device by operating the computer as each part (software element) of the OCT image processing device, and a computer-readable recording medium on which the program is recorded, also fall within the scope of the present invention. [Effects of the Invention]

[0011] According to one aspect of the present invention, a technique can be realized to remove speckles appearing in OCT images from a small number of OCT images. [Brief explanation of the drawing]

[0012] [Figure 1]This is a block diagram showing the configuration of the OCT image processing device 1 according to Embodiment 1 of the present invention. [Figure 2] This is a flowchart showing the flow of the OCT image processing method S1 according to Embodiment 1 of the present invention. [Figure 3] This flowchart shows the flow of OCT image processing method S2 when correcting complex interferometric image data. [Figure 4] This is a schematic diagram illustrating the spatial configuration of complex interferometric image data and an example of its extraction. [Figure 5] This is a schematic diagram illustrating the steps from acquiring 2D complex interferometric image data to the image processing step. [Figure 6] This is a schematic diagram illustrating the training data for a machine model and the steps involved in training the machine model using that data. [Figure 7] This figure shows how the speckle of an OCT image changes depending on the number of SCCP-processed data points. [Modes for carrying out the invention]

[0013] [Embodiment 1] (OCT image processing device 1) Hereinafter, an OCT image processing device 1 according to one embodiment of the present invention will be described in detail with reference to the drawings. The OCT image processing device 1 is a device that performs speckle removal (reduction) processing on images captured using OCT (Optical Coherence Tomography) technology. In other words, the OCT image processing device 1 is not the optical coherence tomography (OCT) system itself that generates OCT images, but a device that reduces speckle in OCT images captured by the OCT system. The OCT image processing device 1 may be provided as an attachment to the optical coherence tomography system, but in the following description, it will be described as a device independent of the optical coherence tomography system. Figure 1 is a block diagram showing the configuration of the OCT image processing device 1.

[0014] As shown in FIG. 1, the OCT image processing apparatus 1 includes a control unit 10, a processor 20, a memory 30, and an input / output IF (interface) 40. The control unit 10 includes an acquisition unit 11, an SCCP processing unit 12, an averaging unit 13, and an imaging unit 14.

[0015] The acquisition unit 11 acquires complex interference image data of the measurement target obtained by irradiating the measurement target with coherent incident light. The complex interference image data is interference signal data acquired by an optical coherence tomography device, and is hereinafter also simply referred to as "OCT image data". The coherent incident light is light with aligned phases directed at the measurement target. In an optical coherence tomography device, the coherent incident light directed at the measurement target interferes with the reflected light reflected from the measurement target, and the interference light is measured by a photosensor and data-processed to be acquired as complex interference image data. The acquisition unit 11 acquires this complex interference image data and records it in the memory 30. The acquisition unit 11 may acquire this complex interference image data from the memory of an arbitrary optical coherence tomography device, or may acquire complex interference image data pre-recorded in various databases or the like.

[0016] The SCCP processing unit 12 repeatedly calculates the product (shifted-conjugate product) of the complex interference image data and the complex conjugate image of the shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined amount, changing the shift amount multiple times. Hereinafter, the "product of the complex interference image data and the complex conjugate image of the shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined amount" will simply be referred to as the "shifted-conjugate product". The number of repetitions is arbitrary. SCCP (Shifted-Complex-Conjugate-Product) processing is the process of deriving the shifted-conjugate product. There may be one or several original complex interference image data from which speckles are to be removed, but the following explanation will describe the case where one complex interference image data is used. This complex interference image data is shifted in a predetermined direction by a predetermined shift amount to obtain shifted image data. The predetermined direction and shift amount can be arbitrarily set by the user. Then, the shifted-conjugate product of the original complex interference image data and the complex conjugate image of the shifted image data is derived. This process is repeated multiple times, varying the shift amount. The process can also be repeated by changing the direction of the shift. SCCP processing refers to the operation of taking the complex conjugate product of the complex interference image data, represented by a complex equation consisting of real and imaginary parts, and the complex conjugate image of the shifted image data. The details of the theoretical principles behind this process will be described later.

[0017] The averaging unit 13 performs a process of averaging the real part data of multiple shift conjugate products. As will be described later, in the absence of aberrations, the element of the real part of the shift conjugate product corresponding to speckle (the part (4-2) in equation (4)) is modulated by shifting. Therefore, by averaging multiple shift images, the speckle is smoothed out and becomes less noticeable. On the other hand, the non-speckle element of the real part (the part corresponding to the incoherent image of the sample under measurement, the part (4-1) in equation (4)) is not modulated even when the complex interference image data is shifted, so the image intensity does not change even when the real part data is averaged. Therefore, overall, the speckle is reduced and the resolution of the OCT image can be improved.

[0018] The imaging unit 14 images the averaged real part data and outputs it to the outside via the input / output IF 40. The output destination may be various display devices, printing devices, memories, etc.

[0019] The processor 20 can be configured using a general-purpose processor such as at least one MPU (Micro Processing Unit) or CPU (Central Processing Unit). Further, the processor 20 may include a dedicated processor configured by an ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array), or PLD (Programmable Logic Device), etc.

[0020] The memory 30 may include multiple types of memories such as ROM (Read Only Memory) and RAM (Random Access Memory). Further, the memory 30 may include a built-in or external memory such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). Also, the memory 30 may be an externally arranged database or the like. As an example, the processor 20 realizes the functions as the acquisition unit 11, SCCP processing unit 12, averaging unit 13, and imaging unit 14 by expanding and executing various control programs recorded in the ROM of the memory 30 in the RAM.

[0021] The input / output IF40 is an interface for sending and receiving data to and from the outside. Communication between the input / output IF40 and the outside may be performed, for example, via the internet. The input / output IF40 may be equipped with a short-range communication device such as WiFi® or Bluetooth® that can connect wirelessly to an internet connection point. Alternatively, it may be a wired connection interface such as a USB connector. For example, the acquisition unit 11 may acquire complex interferometry image data from hospitals, ophthalmology clinics, research facilities, etc., in various locations via the internet and the input / output IF40. Then, it may return the complex interferometry image data with reduced speckle to those hospitals, ophthalmology clinics, research facilities, etc. In this case, the OCT image processing device 1 functions as an OCT image processing platform and can provide a service to remove speckle from OCT images taken at contracted facilities.

[0022] (OCT image processing method S1) Next, we will describe the OCT image processing method S1 performed using the OCT image processing device 1. Figure 2 is a flowchart showing the flow of the OCT image processing method S1. As shown in Figure 2, the OCT image processing method S1 includes steps S11 to S13.

[0023] Step S11 is an SCCP processing step in which the shift conjugate product is calculated multiple times by changing the shift amount. This product is the complex interference image data of the object to be measured obtained by irradiating the object with coherent incident light, and the complex conjugate image of the shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount.

[0024] SCCP processing is performed by the processor 20 of the OCT image processing device 1, specifically the acquisition unit 11 and the SCCP processing unit 12. The complex interference image data, shift image data, and their complex conjugate products are as described above. The details of the method for generating shift image data are described below.

[0025] Figure 4 is a schematic diagram illustrating an example of the spatial structure of OCT image data and its extraction. The rectangular parallelepiped shown at 401 in Figure 4 is the three-dimensional OCT image data space acquired by an optical coherence tomography (OCT) scanner. In this diagram, it is assumed that coherent incident light is irradiated from top to bottom along the z-axis toward the object to be measured. The x and y axes, which are orthogonal to the z-axis and mutually orthogonal, are set as shown in the diagram. From this OCT image data space, the acquisition unit 11 acquires one-dimensional or two-dimensional OCT image data in an arbitrary direction, and the SCCP processing unit 12 shifts it by an arbitrary specified amount in an arbitrary specified direction (one-dimensional, two-dimensional, or three-dimensional direction) to generate shifted image data. The SCCP processing unit 12 then derives the shift conjugate product of the original OCT image data and the shifted image data.

[0026] For example, the dotted line P shown in 401 is a line segment parallel to the x-axis. The acquisition unit 11 acquires one-dimensional OCT image data in a direction parallel to the dotted line P, and the SCCP processing unit 12 can shift it in a direction along the dotted line P to generate one-dimensional shifted image data. Similarly, one-dimensional OCT image data parallel to the dotted line Q parallel to the y-axis, and one-dimensional OCT image data along the dotted line R parallel to the z-axis may be shifted to generate shifted image data. The direction of shifting may be the positive or negative direction of each axis. Also, the direction in which the one-dimensional data is acquired does not need to be parallel to the dotted lines P, Q, and R, but may be any three-dimensional direction.

[0027] For example, the SCCP processing in this case may be a process that repeatedly calculates the shift conjugate product, which is the product of one-dimensional data from the complex interference image data in a direction orthogonal to the direction of incident light and the complex conjugate image of the shifted image data generated by shifting the one-dimensional data by a predetermined shift amount in a direction orthogonal to the direction of incident light, with varying shift amounts.

[0028] Furthermore, the SCCP processing in this case may also be a process that repeatedly calculates the shift conjugate product, which is the product of the one-dimensional data in the direction along the direction of the incident light from the complex interference image data and the complex conjugate image of the shifted image data generated by shifting the one-dimensional data by a predetermined shift amount in the direction along the direction of the incident light, while changing the shift amount.

[0029] Furthermore, when generating shifted image data by shifting one-dimensional OCT image data and deriving the shift conjugate product, it is necessary to shift all the one-dimensional OCT image data that constitute the desired tomographic image and derive the shift conjugate product. This method is computationally more complex than the method for deriving the shift conjugate product of the two-dimensional OCT image plane described below, but this method is also possible.

[0030] Figure 402 shows a two-dimensional OCT image plane (xy plane) OP perpendicular to the z-axis. The acquisition unit 11 may acquire two-dimensional OCT image data on such a plane. A plane image perpendicular to the z-axis is also called an en-face image (a tomographic image facing the direction of irradiation). Alternatively, the acquisition unit 11 may acquire two-dimensional OCT image data such as an OCT image plane PP parallel to the z-axis, as shown in 403. Or, the acquisition unit 11 may acquire two-dimensional OCT image data in any direction in the OCT image space. The SCCP processing unit 12 can shift the acquired two-dimensional OCT image data by an arbitrary amount in the two-dimensional direction within that plane, or in a three-dimensional direction including the direction perpendicular to that plane, and derive the shift conjugate product.

[0031] For example, the SCCP processing in this case may involve generating shifted image data by shifting the complex interference image data at a predetermined depth in at least one of the following directions: the Z direction, which is the depth direction along the direction of the incident light; the X direction, which is perpendicular to the Z direction; and the Y direction, which is perpendicular to both the Z and X directions. The process of calculating the shift conjugate product, which is the product of the original image and the complex conjugate image of these shifted image data, may be repeated multiple times with varying amounts of shift.

[0032] Furthermore, the SCCP processing in this case may involve generating shifted image data by shifting complex interferometric image data on the XY plane at a predetermined depth in a one-dimensional or two-dimensional direction, and then repeating the process of calculating the shift conjugate product, which is the product of the original image and the complex conjugate image of these shifted image data, multiple times with varying amounts of shift.

[0033] In the above embodiment, the shift conjugate product of complex interference image data (original image data) in any direction of one or two dimensions and its shifted image data was derived. However, it is also possible to generate shifted image data shifted in a certain direction from the original image data, and counter-shifted image data shifted by an arbitrary amount in the opposite direction to that direction, and then derive the shift conjugate product of these shifted image data and counter-shifted image data. In other words, the SCCP process may be a process that generates counter-shifted image data by shifting the complex interference image data by an arbitrary amount in the direction opposite to the predetermined direction in which the shifted image data was generated, and then calculates the shift conjugate product of the complex conjugate image of the counter-shifted image data and the complex conjugate image of the shifted image data, repeating this process multiple times with different shift amounts. Alternatively, it may be a process that derives the shift conjugate product of shifted image data shifted in a certain direction and counter-shifted data shifted in a different direction, repeating this process multiple times with different shift amounts.

[0034] The acquired OCT image data contains various measurement errors. Therefore, a correction process may be performed to correct these measurement errors before SCCP processing. This correction process will be described later.

[0035] Returning to Figure 2, step S12 is an averaging step in which the real part data of multiple shift conjugate products are averaged. Step S13 is an imaging step in which the averaged real part data is converted into an image and output.

[0036] The steps from acquiring OCT image data to image processing will be explained with reference to Figure 5. Figure 5 is a schematic diagram showing the steps from acquiring two-dimensional OCT image data to image processing. This schematic diagram shows the case where shifted image data and inversely shifted image data are generated from the original image data, and the shift conjugate product is derived from these.

[0037] A-0 in the upper left of Figure 5 is the 3D OCT image space explained in Figure 4. The following A-1 is a schematic diagram of the acquisition step in which 2D OCT image data S (hereinafter referred to as "original image S") orthogonal to the z axis is acquired from the 3D OCT image space. Focus correction and / or dispersion correction, which will be described later, may be performed on the original image S. Next, shifted image data of the original image S is generated. In the illustrated example, two shifted image data S1 and S2 are generated. Shifted image data S1 shown in A-2 is shifted image data obtained by shifting the original image S by Δx / 2 in the negative x-axis direction and by Δy / 2 in the negative y-axis direction. On the other hand, shifted image data S2 shown in A-3 is shifted image data obtained by shifting the original image S by Δx / 2 in the positive x-axis direction and by Δy / 2 in the positive y-axis direction.

[0038] Next, the complex conjugate S2 of the shifted image data S2. * (A-4) Next, the shifted image data S1 and its complex conjugate S2 * The product of (S1S2 * (B-1) is generated. The shifted image is generated and its complex conjugate product S1S2 * The process of deriving the complex conjugate product, or the complex conjugate product itself, is abbreviated as SCCP. Next, the complex conjugate product S1S2 * The real part is obtained. This step is repeated with a different shift amount. The shift direction may also be changed. The real parts of the multiple complex conjugate products obtained in this way are averaged and the result is converted into an image (D-1). This process yields a corrected image with reduced speckle from the original image S.

[0039] (SCCP theory) Next, we will explain the theoretical aspects of the SCCP processing described above. Optical coherence tomography (OCT) images are composed of complex signals with real and imaginary parts. In image observation, the absolute square intensity of the complex signal is frequently calculated and displayed on a dB scale. On the other hand, the phase of the complex signal is used in phase-sensitive OCT measurements such as Doppler OCT and high-sensitivity displacement measurements. In other words, in such conventional OCT image formation processes, the complex signal is decomposed into amplitude and phase.

[0040] Tomita et al. (K. Tomita et.al., Biomed. Opt. Express 14, 3100 (2023)) demonstrated a novel formulation of OCT, which describes the OCT signal as the sum of two mathematical entities: a "meaningful OCT image" and "speckles." Furthermore, they used this formulation to design a novel complex image processing method for volume difference imaging, which utilizes the "imaginary part" of the manipulated complex OCT signal.

[0041] In this embodiment, a complex image processing method designed based on the formulation by Tomita et al. is used. The image processing method combines the formation of a new complex image with the acquisition of its real and imaginary parts. In real part imaging in this embodiment, speckle reduction is possible using only one original image.

[0042] In the theoretical framework formulated by Tomita et al., the object being measured is modeled as a spatially slowly changing refractive index distribution, and the scatterers are spatially randomly dispersed. This representation of the object being measured is expressed as the "dispersive scatterer model" (DSM). Let (x,y,z) be the position within the object being measured, x and y be the lateral positions, and z be the depth position in the direction of illumination. The spatially slowly changing refractive index is represented by n(x,y,z), and it is assumed that small scatterers are dispersed (i.e., embedded), and that the size of the scatterers is significantly smaller than the spatial variation of n(x,y,z). The position of the i-th scatterer is (xi,yi,zi), and the refractive index of all scatterers is assumed to be the same at ns. The illumination light is incident from above the object being measured, and we consider an en-face image in which only scattering from N scatterers within the coherence gate, which is the measurement surface (the xy plane at depth z0), contributes to the imaging.

[0043] Based on the above assumption, the complex OCT image signal s in a coherence gate at depth z0 is expressed by equation (1) below. This corresponds to the 2D OCT image data S(A-1) in Figure 5. In the following formulas and explanations, variables representing vectors are written in bold, and variables representing scalars are written in regular font.

number

[0044] x and z represent positions in the measurement space, while x' and z' represent positions in the image space. Φi is the phase offset caused by the depth position of the i-th scatterer relative to the central depth of the coherence gate z0. φ is the phase of the complex point response function. Pa is the amplitude of the point response function.

[0045] We assume that Pa, which is included in this complex OCT image signal s, is a Gaussian point response function represented by equation (2) below.

number

[0046] Based on these premises, the product of the shifted image data obtained by shifting image data S by -Δx / 2 in the x-axis direction and -Δy / 2 in the y-axis direction, as explained in Figure 5, and the complex conjugate of the shifted image data obtained by shifting image data S by +Δx / 2 in the x-axis direction and +Δy / 2 in the y-axis direction, and (B-1 in Figure 5) is expressed by the following equation (3).

number

[0047] In equation (3), D and D' are non-binary scatter density maps in 2D and 4D space, respectively. k0 is the wavenumber of the irradiated light, and n is the refractive index of the sample around the scatter. w is the radius of the irradiated light, and w0 is the beam radius at the depth of focus, i.e., the beam waist radius (corresponding to the resolution). R is the phase curvature radius of the wavefront caused by defocusing. RHS means "right side".

[0048] Next, extracting the real part data from equation (3) yields equation (4) below.

number

[0049] Here, Φ pq is a substantially random phase defined by the depth position of the scatterer. In this equation, consider the case where w=w0, that is, the image at the in-focus position. In this case, R approaches infinity, so the cos term in the part of equation (4-1) (the actual image part, not speckle) becomes approximately 1. Also, the other terms become constant terms,

number

number

number

[0050] (OCT image data correction) OCT image data contains data such as focus shift. For example, the focal point of incident coherent light is fixed at a predetermined depth. Therefore, focus shift occurs at depths before and after this point. In the case of a frontal image at a depth shifted from the focal point, focus correction processing can be performed before SCCP processing. For example, the acquisition unit 11 may further perform data correction processing to remove at least a portion of at least one of the low-order aberrations and higher-order aberrations included in the complex interference image data. Low-order aberrations are so-called focus shift aberrations. Higher-order aberrations are, for example, astigmatism and spherical aberration. Data correction processing can be performed, for example, by computational defocusing, and specifically, data correction processing can be performed by phase filtering, complex filtering, data point resampling (ISAM, Interferometric Synthetic Aperture Microscopy), or a combination thereof.

[0051] Furthermore, the acquisition unit 11 may further perform dispersion correction processing to pre-correct the wavelength dispersion included in the complex interference image data. Dispersion correction processing is effective when acquiring shift image data in the Z-axis (depth) direction. Dispersion correction processing can be performed by signal processing methods, hardware methods, or a combination thereof. Specifically, signal processing methods include, for example, a method of multiplying the spectral interference signal by a phase function and then reconstructing the OCT signal (FFT). Hardware methods include, for example, a method of designing the optical system so that the wavelength dispersion of the two arms of the interferometer is equal.

[0052] (Subpixel shift) Next, let's discuss subpixel shifting. Subpixel shifting refers to the process of shifting the original image data by an amount smaller than the size of one pixel when generating shifted image data. In OCT images, one sensor element of the photosensor corresponds to one pixel. Therefore, the shift is usually performed in units of pixel size, but the amount of shift may be smaller than the size of one pixel in complex interferometry image data.

[0053] When shifting by a size smaller than one pixel, it is preferable to create some kind of interpolation data when shifting complex interferometric image data. For example, the pixel density of the image may be improved by linear and / or nonlinear interpolation methods for the complex interferometric image data. For example, the pixel density of the image may be improved by a Fourier domain zero-filling method for the complex interferometric image data before shifting. This allows for accurate reduction of speckle even when sub-pixel shifting is performed.

[0054] As described above, the OCT image processing method S2 for correcting complex interferometric image data will be explained. Figure 3 is a flowchart showing the flow of the OCT image processing method S2. As shown in Figure 3, the OCT image processing method S2 includes steps S21 to S27. The OCT image processing method S2 can be executed by each part of the control unit 10 of the OCT image processing device 1.

[0055] Step S21 is an acquisition step in which complex interference image data of the object to be measured is obtained by irradiating the object with coherent incident light. This acquisition step is as described in the function of the acquisition unit 11.

[0056] Step S22 is a step to remove at least a portion of at least one of the low-order aberrations and higher-order aberrations that may be contained in the complex interference image data. Step S23 is a step to correct image blurring due to wavelength dispersion that may be contained in the complex interference image data. Note that the order in which steps S22 and S23 are performed does not matter, and only a part of them may be performed.

[0057] Step S24 is a step in which the process of deriving the shift conjugate product of complex interference image data and shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount is repeated multiple times, with different shift amounts. The specific method is as described above.

[0058] Step S25 is a step in which, when shifting the complex interferometric image data in step S24, the pixel density of the image is improved for the complex interferometric image data by linear and / or nonlinear interpolation methods. These specific methods are as described above.

[0059] Steps S26 and S27 are the same as steps S12 and S3 of the OCT image processing method S1 described above.

[0060] According to the above-described OCT image processing device 1 or OCT image processing methods S1 and S2, speckles appearing in OCT images can be removed using fewer OCT images (basically just one).

[0061] (Training data for machine models and trained machine models) Next, we will describe the training data for the machine model and the OCT image recognition machine model trained using that training data. Figure 6 is a schematic diagram showing the training data 61 for the machine model and the steps of training an untrained machine model 62 using that training data 61. The training data 61 includes multiple datasets, each consisting of a set of speckle-reduced OCT image data and a label attached to that OCT image data. The speckle-reduced OCT image data is OCT image data obtained by the OCT image processing device or OCT image processing method described above. The label is a so-called ground truth label that indicates what features the OCT image data possesses. The label may be, for example, the name of a disease related to the eye or other organs. Alternatively, the label may be the name of a cell indicating what type of cell it is. The machine model can be any program model, such as a convolutional neural network (CNN) model suitable for image recognition.

[0062] By training an untrained machine model 62 using such training data 61, a trained OCT image recognition machine model 63 capable of accurately identifying and determining the features described in the labels can be generated. Furthermore, by using images with reduced speckle, a machine model capable of recognizing OCT images with even greater accuracy can be generated.

[0063] [Examples of implementation using software] The functions of the OCT image processing device 1 (hereinafter referred to as "the device") are programs that cause the device to function as a computer, and these programs can be realized by programs that cause the computer to function as each control block of the device (particularly each part included in the control unit 10).

[0064] In this case, the device includes a computer having at least one control device (e.g., a processor 20) and at least one storage device (e.g., memory 30) as hardware for executing the program. By executing the program using this control device and storage device, each of the functions described in the above embodiment is realized.

[0065] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0066] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0067] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention. [Examples]

[0068] Next, an embodiment of the present invention (an embodiment in which the speckle of an actual OCT image is reduced) will be described below. Figure 7 is a diagram showing the change in speckle of an OCT image depending on the number of SCCP-processed data. 701 in Figure 7 is the original OCT image (raw image) acquired by an optical coherence tomography (OCT) system. 702 is an averaged image obtained by averaging the real part data of a total of 7 × 7 = 49 shift conjugate products of images shifted by 1 pixel in the x direction from -3 pixels to +3 pixels (including zero pixels) and images shifted by 1 pixel in the y direction from -3 pixels to +3 pixels (including zero pixels).

[0069] Similarly, 703 is an averaged image obtained by averaging the real part data of the shift conjugate product of a total of 11 × 11 = 121 images, which are shifted in the x-direction by 1 pixel increments from -5 pixels to +5 pixels (including zero pixels) and in the y-direction by 1 pixel increments from -5 pixels to +5 pixels (including zero pixels).

[0070] Image 704 is an averaged image obtained by averaging the real part data of the shift conjugate product of a total of 15 × 15 = 225 images, which are shifted in the x-direction by 1 pixel increments from -7 pixels to +7 pixels (including zero pixels) and in the y-direction by 1 pixel increments from -7 pixels to +7 pixels (including zero pixels).

[0071] Focusing on the elliptical areas shown in each figure, we can see that as the number of averaged shift conjugation products increases, the tiny black spots (speckles) within the ellipse disappear, and the outlines of the cells (dark shadow areas) gradually become clearer. Furthermore, the arrows in figures 703 and 704 indicate regions where cell loss has become apparent. Thus, averaging multiple shift conjugation products can reduce speckles, resulting in a clearer image of real cells.

[0072] 〔summary〕 (Aspect 1) OCT image processing method, wherein at least one processor performs SCCP processing, which involves repeatedly calculating the shift conjugate product multiple times by changing the shift amount, the shift conjugate product being the product of complex interference image data of a measurement target obtained by irradiating the measurement target with coherent incident light and the complex conjugate image of shift image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount; averaging processing, which averages the real part data of multiple shift conjugate products; and imaging processing, which converts the averaged real part data into an image and outputs it.

[0073] (Aspect 2) The OCT image processing method according to Embodiment 1, wherein the SCCP processing is a process of generating shifted image data by shifting the complex interference image data at a predetermined depth in at least one of the following directions: the Z direction, which is the depth direction along the direction of the incident light; the X direction, which is the direction perpendicular to the Z direction; and the Y direction, which is the direction perpendicular to both the Z direction and the X direction; and calculating the shift conjugate product, and repeating this process multiple times with a different shift amount.

[0074] (Aspect 3) The OCT image processing method according to embodiment 2, wherein the SCCP processing is a process of generating shifted image data by shifting the complex interference image data on the XY plane at a predetermined depth in a one-dimensional or two-dimensional direction, and calculating the shift conjugate product, and repeating this process multiple times with a different shift amount.

[0075] (Aspect 4) The OCT image processing method according to embodiment 2, wherein the SCCP processing is a process that generates one-dimensional data from the complex interference image data in a direction orthogonal to the direction of the incident light, and shifted image data generated by shifting the one-dimensional data by a predetermined shift amount in a direction orthogonal to the direction of the incident light, and calculates the shift conjugate product, and repeats this process multiple times with different shift amounts.

[0076] (Aspect 5) The OCT image processing method according to embodiment 2, wherein the SCCP processing is a process that generates one-dimensional data in the direction along the direction of the incident light and shifted image data generated by shifting the one-dimensional data by a predetermined shift amount in the direction along the direction of the incident light, and calculates the shift conjugate product, and repeats this process multiple times with different shift amounts.

[0077] (Aspect 6) The OCT image processing method according to embodiment 2, wherein the SCCP processing is a process of generating counter-shifted image data by shifting the complex interference image data by an arbitrary shift amount in the direction opposite to or different from the predetermined direction in which the shifted image data was generated, and deriving the shift conjugate product of the counter-shifted image data and the shifted image data, and repeating this process multiple times with different shift amounts.

[0078] (Aspect 7) The OCT image processing method according to any one of embodiments 1 to 6, wherein the processor further performs data correction processing to remove at least a portion of at least one of the low-order aberrations and high-order aberrations included in the complex interference image data.

[0079] (Pattern 8) The OCT image processing method according to any one of embodiments 1 to 7, wherein the processor further performs dispersion correction processing to pre-correct the wavelength dispersion included in the complex interference image data.

[0080] (Aspect 9) The OCT image processing method according to any one of embodiments 1 to 8, wherein the shift amount is smaller than one pixel of the complex interference image data or smaller than the resolution.

[0081] (Aspect 10) The OCT image processing method according to any one of embodiments 1 to 9, wherein when shifting the complex interference image data, the processor further performs a process to improve the pixel density of the image by a linear and / or nonlinear interpolation method on the complex interference image data.

[0082] (Aspect 11) Training data for training a machine model that performs OCT image judgment, comprising an OCT image obtained using the OCT image processing method described in any one of embodiments 1 to 10, and a label indicating the characteristics of the OCT image.

[0083] (Aspect 12) An OCT image recognition machine model trained using the training data described in Embodiment 11.

[0084] (Aspect 13) An acquisition unit that acquires complex interference image data of a measurement target obtained by irradiating the measurement target with coherent incident light, An SCCP processing unit derives the shift conjugate product, which is the product of the complex interference image data and the complex conjugate image of the shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount, by repeating the process multiple times with varying shift amounts. An averaging unit that averages the real part data of multiple shift conjugate products, An imaging unit that converts the averaged real data into an image and outputs it, An OCT image processing device equipped with the following features.

[0085] (Aspect 14) An OCT image processing program for causing a computer to function as an OCT image processing apparatus according to embodiment 13, wherein the computer functions as the acquisition unit, the SCCP processing unit, the averaging unit, and the imaging unit.

[0086] (Aspect 15) A computer-readable non-temporary recording medium that records the OCT image processing program described in Embodiment 14. [Explanation of symbols]

[0087] 1…OCT Image Processing Device 10…Control Unit 11…Acquisition part 12…SCCP Processing Unit 13...Averaging section 14…Image Processing Section 20… Processor 30...memory 40…Input / Output Interface 61...Training data 62…Untrained machine models 63…Pre-trained machine model

Claims

1. At least one processor, SCCP processing involves repeatedly calculating the shift conjugate product, which is the product of the complex interference image data of the object to be measured obtained by irradiating the object with coherent incident light, and the complex conjugate image of the shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount, multiple times while changing the shift amount. An averaging process that averages the real part data of multiple shift conjugate products, The aforementioned averaged real data is converted into an image and output as an image, Execute OCT image processing method.

2. The OCT image processing method according to claim 1, wherein the SCCP processing is a process of generating shifted image data by shifting the complex interference image data at a predetermined depth in at least one of the following directions: the Z direction, which is the depth direction along the direction of the incident light; the X direction, which is perpendicular to the Z direction; and the Y direction, which is perpendicular to both the Z direction and the X direction; and calculating the shift conjugate product, and repeating this process multiple times with a different shift amount.

3. The OCT image processing method according to claim 2, wherein the SCCP processing is a process of generating shifted image data obtained by shifting the complex interference image data on the XY plane at a predetermined depth in a one-dimensional or two-dimensional direction, and calculating the shift conjugate product, and repeating this process multiple times with a different shift amount.

4. The OCT image processing method according to claim 2, wherein the SCCP processing is a process of generating one-dimensional data from the complex interference image data in a direction orthogonal to the direction of the incident light, and shifted image data generated by shifting the one-dimensional data by a predetermined shift amount in a direction orthogonal to the direction of the incident light, and calculating the shift conjugate product, and repeating this process multiple times with a different shift amount.

5. The OCT image processing method according to claim 2, wherein the SCCP processing is a process of generating one-dimensional data in the direction along the direction of the incident light and shifted image data generated by shifting the one-dimensional data by a predetermined shift amount in the direction along the direction of the incident light from the complex interference image data, and calculating the shift conjugate product, and repeating the process multiple times with a different shift amount.

6. The OCT image processing method according to claim 2, wherein the SCCP processing is a process of generating counter-shifted image data by shifting the complex interference image data by an arbitrary shift amount in the direction opposite to or different from the predetermined direction in which the shifted image data was generated, and deriving the shift conjugate product of the counter-shifted image data and the shifted image data, and repeating this process multiple times with different shift amounts.

7. The OCT image processing method according to any one of claims 1 to 6, wherein the processor further performs data correction processing to remove at least a portion of at least one of the low-order aberrations and high-order aberrations included in the complex interference image data.

8. The OCT image processing method according to any one of claims 1 to 6, wherein the processor further performs dispersion correction processing to pre-correct the wavelength dispersion included in the complex interference image data.

9. The OCT image processing method according to any one of claims 1 to 6, wherein the shift amount is smaller than one pixel of the complex interference image data or smaller than the resolution.

10. The OCT image processing method according to claim 9, wherein when shifting the complex interference image data, the processor further performs a process to improve the pixel density of the image by a linear and / or nonlinear interpolation method on the complex interference image data.

11. Training data for training a machine model that performs OCT image judgment, comprising an OCT image obtained using the OCT image processing method described in any one of claims 1 to 6, and a label indicating the characteristics of the OCT image.

12. An OCT image recognition machine model trained using the training data described in claim 11.

13. An acquisition unit that acquires complex interference image data of a measurement target obtained by irradiating the measurement target with coherent incident light, An SCCP processing unit derives the shift conjugate product, which is the product of the complex interference image data and the complex conjugate image of the shifted image data generated by shifting the complex interference image data in a predetermined direction by a predetermined shift amount, and repeats this process multiple times while changing the shift amount. An averaging unit that averages the real part data of multiple shift conjugate products, An imaging unit that converts the averaged real data into an image and outputs it, An OCT image processing apparatus equipped with [a specific feature].

14. An OCT image processing program for causing a computer to function as an OCT image processing apparatus according to claim 13, wherein the computer functions as the acquisition unit, the SCCP processing unit, the averaging unit, and the imaging unit.

15. A computer-readable non-temporary recording medium that records the OCT image processing program described in claim 14.

Images