JP2008039651A - Processing method of optical tomographic image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To realize a method for correcting the strain of a three-dimensional image due to the movement of an object to be measured using only three-dimensional measuring data without using a complicated add-on system such as a compensation optics or a motion track, and a method for automatically extracting a layered structure from the image corrected in strain to separate, measure and display the structure present in the layered structure. <P>SOLUTION: In an optical tomographic imaging method due to optical coherence tomography 1 for constituting a two-dimensional tomographic image parallel to the axis in the depth direction of the object 8 to be measured using optical coherence tomography 1 or acquiring a plurality of images while shifting a position in the direction vertical to the two-dimensional tomographic image to constitute the three-dimensional image, the Doppler signal caused by the movement shift of the object 8j to be measured is detected and the correction of the positional shift of the two-dimensional tomographic image is performed on the basis of this detection result to reconstitute the three-dimensional image to correct the strain of the optical tomographic image of optical coherence tomography due to the movement of the object 8 to be measured. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  A three-dimensional tomographic image obtained by optical coherence tomography includes all three-dimensional information of an object to be measured. The present invention relates to a method for processing an optical tomographic image for accurately extracting the information.

  One of the non-destructive tomographic techniques used in the medical field or the like is an optical tomographic imaging method “optical coherence tomography” (OCT) (see Patent Document 1). Since OCT uses light as a measurement probe, it has the advantage that it can measure the refractive index distribution, spectral information, polarization information (birefringence distribution), etc. of the measured object.

  The basic OCT 43 is based on a Michelson interferometer, and its principle will be described with reference to FIG. The light emitted from the light source 44 is collimated by the collimator lens 45 and then divided into reference light and object light by the beam splitter 46. The object light is condensed on the measurement object 48 by the objective lens 47 in the object arm, scattered and reflected there, and then returns to the objective lens 47 and the beam splitter 46 again.

  On the other hand, the reference light passes through the objective lens 49 in the reference arm, is reflected by the reference mirror 50, and returns to the beam splitter 46 through the objective lens 49 again. The object light and the reference light that have returned to the beam splitter 46 in this way are incident on the condensing lens 51 together with the object light and are collected on the photodetector 52 (photodiode or the like).

  The light source 44 of the OCT uses a light source of light having low temporal coherence (light emitted from the light source at different times is extremely difficult to interfere with each other). In a Michelson interferometer using temporally low coherence light as a light source, an interference signal appears only when the distance between the reference arm and the object arm is approximately equal. As a result, when the intensity of the interference signal is measured by the photodetector 52 while changing the optical path length difference (τ) between the reference arm and the object arm, an interference signal (interferogram) for the optical path length difference is obtained.

  The shape of the interferogram shows the reflectance distribution in the depth direction of the measurement object 48, and the structure in the depth direction of the measurement object 48 can be obtained by one-dimensional axial scanning. Thus, in the OCT 43, the structure in the depth direction of the measurement object 48 can be measured by optical path length scanning.

  In addition to the scanning in the axial direction, a two-dimensional cross-sectional image of the object to be measured can be obtained by performing a two-dimensional scanning by adding a horizontal mechanical scanning. The scanning device that performs the horizontal scanning includes a configuration in which the object to be measured is directly moved, a configuration in which the objective lens is shifted while the object is fixed, and a pupil of the objective lens while the object to be measured and the objective lens are fixed. The structure etc. which rotate the angle of the galvanometer mirror in the surface vicinity are used.

  As a development of the above basic OCT, a wavelength scanning OCT (Swept Source OCT, abbreviated as “SS-OCT” for short) that scans the wavelength of a light source to obtain a spectrum interference signal, and a spectroscope are used. There is a spectrum domain OCT for obtaining a spectrum signal. The latter includes Fourier domain OCT (Fourier Domain OCT, abbreviated as “FD-OCT”; see Patent Document 2), and polarization-sensitive OCT (Polarization-Sensitive OCT, abbreviated as “PS”. -OCT "(see Patent Document 3).

  The wavelength scanning type OCT obtains a three-dimensional optical tomographic image by changing the wavelength of a light source by a high-speed wavelength scanning laser, rearranging interference signals using a light source scanning signal acquired in synchronization with a spectrum signal, and applying signal processing. Is. As a means for changing the wavelength of the light source, a device using a monochromator can be used as the wavelength scanning OCT.

  In the Fourier domain OCT, the wavelength spectrum of the reflected light from the object to be measured is acquired with a spectrometer (spectrum spectrometer), and Fourier transform is performed on this spectrum intensity distribution, so that the real space (OCT signal space) is obtained. This Fourier domain OCT does not need to scan in the depth direction, and can measure the cross-sectional structure of the object to be measured by scanning in the x-axis direction.

  Like the Fourier domain OCT, the polarization-sensitive OCT acquires the wavelength spectrum of the reflected light from the object to be measured with a spectrum spectrometer. For example, horizontal linearly polarized light, vertical linearly polarized light, 45 ° linearly polarized light, and circularly polarized light passed through a four-wavelength plate, etc. Only the horizontally polarized component is incident on the spectrum spectrometer to cause interference, and only the component having a specific polarization state of the object light is extracted and subjected to Fourier transform. This polarization sensitive OCT also does not need to be scanned in the depth direction.

Doppler OCT is a method for obtaining the velocity of blood flow or the like by utilizing the fact that the amount of phase change obtained by Fourier transform of spectral interference information corresponds to the moving velocity of an object to be measured as a Doppler signal. And can be summarized in Fourier domain OCT (see Non-Patent Document 1).
JP 2002-310897 A JP 11-325849 A JP 2004-028970 A BR White et al., Optics Express, Vol. 11, No. 25 (2003), p. 3490

  By the way, when measuring the shape of a living body three-dimensionally, if the object to be measured moves and the relative positional relationship between the object to be measured and the measuring instrument moves within the measurement time, the image is distorted due to the movement of the object to be measured. Has occurred.

  This not only distorts the spatial information of the object to be measured, but also causes an error in the measured value when measuring the flow velocity or the like.

  Conventionally, as a means for solving the problem, correction of a measurement position is mainly performed in real time using an adaptive optical system, a motion track (means for sequentially measuring movement), or the like. However, the compensation optical system and the motion track are extremely complicated means, and it is expensive and troublesome to attach these means.

  That is, the conventionally used adaptive optics system uses a variable-shape mirror (the surface shape of the mirror is deformed by computer control etc.) to monitor the reflected image and the like so that its shape and position do not change during measurement. The optical system is changed so as to compensate for the movement of the object to be measured, and a deformable mirror, a liquid crystal optical element, a prism, or the like is used.

  Measured object monitoring measurement system, optical system compensation amount calculation, optical system deformation, post-deformation monitor object measurement, optical system compensation amount calculation, optical system deformation feedback loop The system is complicated and the optical system is expensive. There is no guarantee of stability.

  A motion track is a simplified version of the above-mentioned adaptive optics system that monitors a specific point of the object to be measured and monitors the movement of the object to be measured from its movement, which also requires a separate optical system and processing system. It is.

  The object of the present invention is to solve the above-described conventional problems, and as a means for solving this problem, an object to be measured is used by using only three-dimensional measurement data without using a complicated additional system such as adaptive optics or a motion track. In particular, the initial relative position and distortion between two-dimensional tomographic images composed of a plurality of one-dimensional images measured at the same time can be determined between the images. The means for correcting by the Doppler information representing the movement displacement is realized.

  By the way, it is generally difficult to extract useful information from an optical tomographic image. It takes a lot of experience and skill to obtain the information of interest from a lot of noise and fine structures. Only possible with careful observation by a specialist in the field. Alternatively, it is possible to extract information by using a huge database and comparing patterns with it, but in order to do so, a supercomputer and a large-capacity memory are required.

  In screening, such as group health examinations that handle a large amount of data, it is necessary to extract information that is automatically obtained. In view of such a demand, the present invention realizes a method of completely automatically separating a three-dimensional structure in the depth direction.

  In order to solve the above-described problems, the present invention uses optical coherence tomography, and uses a plurality of one-dimensional tomographic images while shifting the position in the vertical direction (B scan direction) to the depth axis (A scan axis) of the object to be measured. To obtain a two-dimensional tomographic image (B-scan image) parallel to the axis in the depth direction of the object to be measured, and a plurality of 2 while shifting the position vertically to the two-dimensional tomographic image (C-scan direction). In an optical tomography method using optical coherence tomography that acquires a three-dimensional tomographic image and constructs a three-dimensional image, positional deviations of the plurality of one-dimensional tomographic images caused by movement blur of the measured object are used using Doppler signal information. And detecting the position of the one-dimensional tomographic image based on the detection result, reconstructing the two-dimensional tomographic image, and performing optical coherence tomography. It provides a method of processing an optical tomographic image and correcting distortion of the image due to movement of the object to be measured in the optical tomographic imaging method.

  In the optical tomographic image processing method, a two-dimensional tomographic image (B-scan image) parallel to the axis in the depth direction (A-scan axis) of the measured object is perpendicular to the two-dimensional tomographic image using optical coherence tomography. In the optical tomographic imaging method using optical coherence tomography that forms a three-dimensional image by shifting a plurality of positions in the direction (C scan direction), each of the plurality of two-dimensional tomographic images that are acquired due to movement blur of the object to be measured The positional deviation of the image is detected using Doppler signal information, the positional deviation of the two-dimensional tomographic image is corrected based on the detection result, the three-dimensional image is reconstructed, and the optical tomographic image of optical coherence tomography It is preferable to correct the distortion of the image due to the movement of the measurement object in the conversion method.

  In the optical tomographic image processing method, the optical coherence tomography is Doppler optical coherence tomography, and in the Doppler optical coherence tomography, a velocity due to movement blur that occurs when a flow velocity or a moving amount of a measurement target in a tomographic image is obtained. It is preferable to correct image distortion due to movement of the measurement object in the optical tomographic imaging method of optical coherence tomography, which is used to correct the measurement error.

  In the optical tomographic image processing method, the Doppler signal represents the moving speed of the object to be examined and is represented by the amount of change in the phase component of the optical coherence tomography image, and the A scan at different positions in the B scan direction. Obtaining a histogram of Doppler signals between data, regarding the most frequent component in the histogram as moving blur, and using the value of the component, an image by the movement of the measurement object in the optical tomography of optical coherence tomography It is preferable to correct the distortion.

  In the optical tomographic image processing method, the median value of the Doppler signal is assumed to be a movement blur, and the median value is used to correct image distortion due to the movement of the object to be measured in the optical tomographic imaging method of optical coherence tomography. preferable.

  In the optical tomographic image processing method, it is preferable that the optical tomographic image processing method separates the distortion-corrected image when a different structure exists in the depth direction of the object to be measured.

  In the optical tomographic image processing method, in the optical tomographic image processing method, it is preferable that the structure in the separated layer structure is displayed separately from the structures of other layer structures.

  Since the method of the present invention is as described above, the movement of the object to be measured is performed using only the three-dimensional measurement data without using a complicated, expensive, and cumbersome additional system such as adaptive optics and a motion track. It is possible to correct the distortion of the derived image.

  An optical tomographic image processing method according to the present invention will be described below with reference to the drawings based on first to third embodiments, particularly the best mode for carrying out a correction method in optical tomographic imaging of optical coherence tomography. .

  FIG. 1 is a diagram illustrating an overall configuration of the FD-OCT 1. A broadband light source 2, a low coherence interferometer 3, and a spectrometer 4 (spectrometer) are provided. Since the FD-OCT 1 obtains resolution in the depth direction using the principle of low coherence interference, a broadband light source 2 such as an SLD (super luminescent diode) or an ultrashort pulse laser is used as a light source.

  The light emitted from the broadband light source 2 is first split into object light and reference light by the beam splitter 5. Of these, the object light is reflected by the galvanometer mirror 7 through the lens 6 and irradiates the measurement object 8 (biological sample such as the fundus), and is reflected and scattered there and then guided to the spectrometer 4. On the other hand, the reference light is reflected by the reference mirror 10 (plane mirror) through the lens 9 and then guided to the spectroscope 4 in parallel with the object light. These two lights are simultaneously dispersed by the diffraction grating 11 of the spectroscope 4 and interfere in the spectral region. As a result, the spectral interference fringes are measured by the CCD 12.

  By performing appropriate signal processing on the spectral interference fringes, it is possible to obtain a differential of the one-dimensional refractive index distribution in the depth direction at a certain point of the measured object 8, that is, a reflectance distribution. Furthermore, a two-dimensional tomographic image (FD-OCT image) can be obtained by driving the galvanometer mirror 7 and one-dimensionally scanning a measurement point on the measurement object 8.

  In normal OCT, in order to obtain a two-dimensional tomographic image, scanning in the depth (optical axis) direction (this scanning is referred to as “A-scan”, which is referred to as “A-direction” and “A-scan direction”). And two-dimensional mechanical scanning is required for vertical operation (this scanning is called “B-scan”, and this direction is also called “B-direction”, “B-scan direction”). In FD-OCT1, since A-scan is unnecessary and backscattering data in the depth direction can be acquired by one measurement, only one-dimensional mechanical scanning of B-scan is required.

  A scan in a direction perpendicular to the plane formed by the A-direction and the B-direction is referred to as “C-scan”, and this direction is also referred to as “C-direction” or “C-scan direction”. In short, in FD-OCT1, two-dimensional scanning (B-scan and C-scan) is performed in the plane, so that high-speed tomographic measurement is possible, and two-dimensional and three-dimensional information inside the measurement object 8 can be obtained. it can.

  In the FD-OCT 1 as described above, the present invention acquires two-dimensional data of the measurement object 8 and detects a Doppler signal in order to correct image distortion caused by the movement of the measurement object 8 itself. This is a method of correcting the movement in the B-direction by integrating the above, and correcting the positional relationship using information of the measurement data itself.

  That is, the present invention uses optical coherence tomography to acquire a plurality of one-dimensional tomographic images while shifting the position in the vertical direction (B scan direction) to the axis in the depth direction (A scan axis) of the object to be measured. A two-dimensional tomographic image (B-scan image) having two sides in the depth direction of the measurement object and a direction perpendicular thereto is acquired, and a plurality of the two-dimensional tomographic images are shifted while being shifted in the vertical direction (C-scan direction). In the optical tomographic imaging method using optical coherence tomography that acquires a two-dimensional tomographic image and forms a three-dimensional image, this is a method of finally correcting the distortion of the three-dimensional image.

  Specifically, the positional deviation of a plurality of one-dimensional tomographic images caused by the movement blur of the measurement object in the A scan direction during the B scan (this is referred to as “bulk motion” in this specification) The detection is performed using information, and the positional deviation of the one-dimensional tomographic image is corrected based on the detection result, the two-dimensional tomographic image is reconstructed, and a plurality of each acquired due to the bulk motion is obtained in the C scan. The positional deviation of the two-dimensional tomographic image is detected using the Doppler signal information, the positional deviation of the two-dimensional tomographic image is corrected based on the detection result, and finally the three-dimensional image is reconstructed. In this method, the distortion of the three-dimensional image due to the movement of the object to be measured is corrected.

  Hereinafter, in the first embodiment of the present invention, a method for correcting the positional deviation of a one-dimensional tomographic image caused by the bulk motion in the A scan direction during the B scan, which is the basis of the present invention, will be described in detail. The correction of the positional deviation of the two-dimensional tomographic image caused by the bulk motion in the A scan direction during the C scan is performed by the same method.

  Alternatively, by using optical coherence tomography, a plurality of two-dimensional tomographic images parallel to the axis in the depth direction of the object to be measured are acquired while shifting the position in the vertical direction with respect to the two-dimensional tomographic image to form a three-dimensional image. In the optical tomographic imaging method by tomography, a positional shift of each of the plurality of acquired two-dimensional tomographic images caused by the movement blur of the measurement object is detected using digital correlation, and the two-dimensional tomographic image is based on the detection result. The distortion of the three-dimensional image due to the movement of the object to be measured in the optical tomographic imaging method of optical coherence tomography, wherein the three-dimensional image is reconstructed, may be corrected.

In FD-OCT1, as shown in FIG. 2, a one-dimensional cross-sectional image obtained in A- scanning at a first lateral position (B-scan direction) and B _i, minute distance from the first position transverse A one-dimensional cross-sectional image obtained by A-scan at the second position moved in the direction is defined as B _{i + 1} . Since these two data B _i and B _{i + 1} are measured at different times, the relative position of the data that should be the same depth is shifted by the movement of the measured object 8 (living body) in the A direction. There is a possibility.

In FD-OCT, when obtaining OCT image data, Fourier transform of spectral interference data is performed. Since it is generally a complex number after Fourier transform, the data can be separated into an intensity component and a phase component. The intensity component of the OCT image data represents the scattering intensity from the object to be measured, but the difference Δφ _{i + 1} (z) = φ _{i + 1} (z) −φ between the phase components of B _i and B _{i + 1} collected at different times. _i (z) is proportional to the amount of movement of the position of B _i and B _{i + 1 in} the A scan direction as a Doppler signal. Here, φ _i is a phase component of B _i , φ _{i + 1} is a phase component of B _{i + 1} , and z is a coordinate in the depth direction along the A-scan. Note that the fact that the image structure is shifted means that the phase is different. This is because, mathematically, a positional shift is extracted as a phase component at the stage of Fourier transform.

Obtains a difference [Delta] [phi _{i + 1} of the phase component of _{B i} and _{B i + 1} for the A-scan data of next to each other (i and i + 1 th B-scan) (z-direction), a histogram as shown in FIG. The image consists of a region where the measured object exists (bulk portion) and a region where the measured object does not exist (background). Since there is no movement in the background area, the phase difference is uniformly distributed from −π to π. Since the bulk portion occupies a large portion, the phase difference component corresponding to the bulk motion becomes the peak of the histogram. The phase change (phase difference) and [Delta] [phi _B.

Or all z arranging the phase difference [Delta] [phi _{i + 1} in order of magnitude for the median value (median: data are arranged in order of size, the value of the data in the middle) may be a phase change due to bulk motion with [Delta] [phi _{i + 1} of the. This phase change is assumed to be Δφ _B.

The relationship between the phase change and the amount of movement (bulk motion speed) ν _{i that} the object to be measured has moved in the measurement time interval (measurement time interval of B _i and B _{i + 1} ) is represented by ν _i = (λ / a 2n · 2πΔT) φ _B. Here, λ is the center wavelength of the light source, n is the refractive index of the object to be measured, and ΔT is the time interval of the A scan.

The bulk motion amount d _z during the B scan is obtained by multiplying the bulk motion speed ν _i by the A scan time interval ΔT and integrating it in the B scan direction. That is, it is shown by the following formula.

(Means for correcting from the amount of bulk motion)
Thus, the bulk motion amount is integrated, and fitting is performed with a polynomial (for example, d _z = a ₀ + a ₁ z (a ₀ , a ₁ is a constant)). Since the movement of the living body during measurement is not so large, a median shift (a ₀ ≠ 0, a ₁ = 0) or a linear function (a ₁ ≠ 0) is effective. The data in the B-scan direction is moved in the A-scan direction so that the shape of this function becomes flat, the bulk motion in the B-scan direction is corrected, and a still image of the measured object can be obtained.

  Alternatively, without using integration and function fitting, the adjacent B scan data is moved in the A scan direction by the amount of the adjacent bulk motion amount, the bulk motion in the B scan direction is corrected, and the still image of the object to be measured is obtained. May be obtained.

  Velocity of blood flow in blood vessels in a two-dimensional image using Doppler OCT (Doppler optical coherence tomography) that measures the velocity of an object to be measured from phase data obtained by Fourier transforming spectral interference signals in FD-OCT When determining the direction, the correction of bulk motion has the effect of stopping the movement of the object to be measured and removing the error of velocity measurement such as blood flow.

  Although the method for correcting the positional deviation of the one-dimensional tomographic image due to the bulk motion in the A scan direction during the B scan has been described in detail, the same method is caused by the bulk motion in the A scan direction during the C scan. By applying the correction to the displacement of the two-dimensional tomographic image, it is possible to finally correct the displacement caused by the bulk motion in the three-dimensional image.

  Alternatively, by using optical coherence tomography, a plurality of two-dimensional tomographic images parallel to the axis in the depth direction of the object to be measured are acquired while shifting the position in the vertical direction with respect to the two-dimensional tomographic image to form a three-dimensional image. In the optical tomographic imaging method by tomography, a positional shift of each of the plurality of acquired two-dimensional tomographic images caused by the movement blur of the measurement object is detected using digital correlation, and the two-dimensional tomographic image is based on the detection result. The distortion of the three-dimensional image due to the movement of the object to be measured in the optical tomographic imaging method of optical coherence tomography, wherein the three-dimensional image is reconstructed, may be corrected.

Specifically, a one-dimensional cross-sectional image obtained by A-scan at the first position in the C-scan direction is C _i, and A at the second position moved laterally from the first position by a minute distance. -Let C _{i + 1} be the one-dimensional cross-sectional image obtained by scanning. Since these two data C _i and C _{i + 1} are measured at different times, the relative position of the data that should originally have the same depth shifts due to the movement of the measured object 8 (living body) in the A direction. There is a possibility.

The difference Δφ _{i + 1} (z) = φ _{i + 1} (z) −φ _i (z) between the phase components of C _i and C _{i + 1} collected at different times is the movement of the position of C _i and C _{i + 1 in} the A scan direction as a Doppler signal It is proportional to the amount. Here, φ _i is a phase component of C _i , φ _{i + 1} is a phase component of C _{i + 1} , and z is a coordinate in the depth direction along the A-scan. A histogram is created by obtaining a difference Δφ _{i + 1} between the phase components of C _i and C _{i + 1} for the A scan data (z direction) of adjacent (i and i + 1th C scan). Since there is no movement in the background region, the phase difference is uniformly distributed from −π to π.

Since the bulk portion occupies a large portion, the phase difference component corresponding to the bulk motion becomes the peak of the histogram. This phase change is assumed to be Δφ _B. Or all z arranging the phase difference [Delta] [phi _{i + 1} in order of magnitude for the median value (median: data are arranged in order of size, the value of the data in the middle) may be a phase change due to bulk motion with [Delta] [phi _{i + 1} of the.

This phase change is assumed to be Δφ _B. The relationship between the phase change Δφ _B and the movement amount (bulk motion speed) ν _{i in which} the measured object is moved at the measurement time interval (measurement time interval of B _i and B _{i + 1} ) is expressed as ν _i = the (λ / 2n · 2πΔT) φ B. Here, λ is the center wavelength of the light source, n is the refractive index of the object to be measured, and ΔT is the time interval of the A scan. The bulk motion amount d _z during the C scan is obtained by multiplying the bulk motion speed ν _i by the A scan time interval ΔT and integrating it in the C scan direction. That is, it is shown by the following formula 1.

Thus, the bulk motion amount is integrated, and fitting is performed with a polynomial (for example, d _z = a ₀ + a ₁ z (a ₀ , a ₁ is a constant)). Since the movement of the living body during measurement is not so large, a median shift (a ₀ ≠ 0, a ₁ = 0) or a linear function (a ₁ ≠ 0) is effective. The data in the C scan direction is moved in the A scan direction so that the shape of this function is flat, the bulk motion in the C scan direction is corrected, and a still image of the measured object can be obtained.

  Alternatively, without using integration and function fitting, the adjacent C scan data is moved in the A scan direction by the amount of the adjacent bulk motion amount, the bulk motion in the C scan direction is corrected, and the still image of the measured object is obtained. May be obtained.

  That is, in the case of three-dimensional image measurement, the same correction is possible for the B scan and the C scan in the vertical direction, and the bulk motion that is the movement of the living body during the acquisition of the three-dimensional image is converted into the two-dimensional B scan and C scan. By correcting in the direction, the distortion of the three-dimensional image can be corrected.

  FIG. 4 is a diagram showing an overall configuration of a PS-FD-OCT 13 (polarization-sensitive spectral interference tomography apparatus) according to Embodiment 2 of the optical tomographic image processing method of the present invention. As in the first embodiment, a broadband light source 2, a low coherence interferometer 3 (Michelson interferometer), and a spectrometer 4 (spectrometer) are provided. A specific configuration will be described below together with the operation.

  The light emitted from the light source 2 is reduced in power by the optical wedge 14 and then becomes horizontal linearly polarized light (hereinafter referred to as “H”) by the polarizer 15. Then, the polarization state of the incident light is converted into horizontal linearly polarized light (H), vertical linearly polarized light (hereinafter referred to as “V”), 45 ° linearly polarized light (hereinafter referred to as “P”) by the ½ wavelength plate 16 and the ¼ wavelength plate 17. ”) And right-handed circularly polarized light (hereinafter referred to as“ R ”), and selectively adjusts the beam into reference light and light incident on the object to be measured 8 by the beam splitter 5.

  The reference light divided by the beam splitter 5 is adjusted by the reference mirror 10 of the reference light optical system and the two quarter-wave plates 18 and 19 so that the polarization state becomes H, V, P, and R. It is incident again on the splitter 5. On the other hand, the light incident on the measurement object 8 is collected by the lens 6 at one point on the measurement object 8 and reflected to the beam splitter 5 as object light. The beam splitter 5 transmits the incident reference light and reflects the object light by 45 °, and superimposes them.

  The reference light adjusted in such a manner that the polarization state emitted from the beam splitter 5 is superimposed as H, V, P, and R and the object light reflected from the measured object 8 are mirror 20. Then, the polarization state is changed to H through the quarter-wave plate 21 and the half-wave plate 22 and enters the spectroscope 4 including the diffraction grating 11, the lens 23, and the CCD 12.

  In this way, by causing the reference light of specific polarization (reference light in any polarization state of H, V, P, or R) to interfere with the object light, only the specific polarization component of the object light causes spectral interference fringes on the CCD 12. As a result, only the component having the same polarization state as the reference light in the object light can be extracted as a signal. Then, this spectral interference fringe is taken into a computer (not shown), one horizontal line is extracted from one point on the y-axis of the image, and a space is obtained by discrete Fourier transform (DFT: Fast Fourier transform). Compute a general Fourier transform.

  Thereby, a one-dimensional correlation signal between the reference beam and the object beam is obtained. Further, by obtaining a Mueller matrix by combining these signal intensities, polarization information inside the object to be measured 8 can be captured.

  In the second embodiment, similarly to the first embodiment, the two-dimensional data of the measured object 8 is acquired, and the Doppler signal is detected in order to correct the distortion of the image resulting from the movement of the measured object 8 itself. By accumulating it, the positional deviation of the one-dimensional tomographic image due to the bulk motion in the A-direction in the B scan is corrected to reconstruct the two-dimensional tomographic image, and a plurality of C scans are acquired in the same manner. 2D tomographic image positional deviation is detected using Doppler signal information, the positional deviation of the 2D tomographic image is corrected based on the detection result, and finally the 3D image is reconstructed. . Since the details are exactly the same as those in the first embodiment, the description thereof is omitted.

  FIG. 5 is a diagram showing an overall configuration of a wavelength scanning OCT 24 to which the correction method according to the third embodiment of the optical tomographic image processing method of the present invention is applied. The output light emitted from the wavelength scanning light source 25 is sent to the fiber coupler 27 through the fiber 26. In the fiber coupler 27, the output light is divided into object light that is irradiated onto the measurement object 29 through the fiber 28 and reference light that is irradiated onto the fixed reference mirror 31 through the fiber 30.

  The object light is irradiated and reflected on the measurement object 29 through the fiber 28, the lens 32, the scanning mirror 33 and the lens 34 having variable angles, and returns to the fiber coupler 27 through the same route. The reference light is irradiated and reflected on the fixed reference mirror 31 through the fiber 30, the lens 35, and the lens 36, and returns to the fiber coupler 27 through the same route.

  Then, the object light and the reference light are overlapped by the fiber coupler 27 and sent to the optical detector 38 (a point sensor such as a PD (photodiode) is used) through the fiber 37 and detected as a spectrum interference signal. Is taken into the computer 39. Based on the detection output of the photodetector 38, cross-sectional images of the measured object 29 in the depth direction (A direction) and the scanning direction of the scanning mirror (B direction) are formed. A display 40 is connected to the computer 39.

  Here, the wavelength scanning light source 25 is a light source that scans while changing the wavelength with time, that is, a light source having a wavelength-dependent wavelength. This makes it possible to obtain the reflectance distribution in the depth direction of the object 29 to be measured and to obtain the structure in the depth direction without scanning (moving, A-scan) the fixed reference mirror 31. Scanning in the primary direction A two-dimensional tomographic image can be formed simply by performing (B-scan).

  As in the first embodiment, the third embodiment acquires two-dimensional data of the measured object 29 and corrects the distortion of the image derived from the movement of the measured object 29 itself. In order to correct the distortion of the image resulting from the motion, the Doppler signal is detected and integrated to correct the positional deviation of the one-dimensional tomographic image caused by the bulk motion in the A-direction in the B scan. A two-dimensional tomographic image is reconstructed. Similarly, a plurality of C-scan acquired two-dimensional tomographic image positions are detected using Doppler signal information, and the position of the two-dimensional tomographic image is determined based on the detection result. Deviation correction is performed, and a three-dimensional image is finally reconstructed. Since the details are exactly the same as those in the first embodiment, the description thereof is omitted.

  In the fourth embodiment, when there is a different structure in the depth direction of the object to be measured in the optical tomographic image whose distortion has been corrected as in the first to third embodiments, a separation method, that is, the A scan direction ( This is a method for detecting the layer structure in the direction of the optical axis).

  For example, it is necessary to extract information that is desired to be automatically obtained in screening such as a group health examination that handles a lot of data. In view of such a demand, the present invention realizes a method of completely automatically separating a three-dimensional structure in the depth direction. In the fourth embodiment, analysis of the fundus layer structure will be described as an example, but the same applies to general images.

  The method of Example 4 will be described. First, an area of interest is extracted. This is based on preliminary knowledge about the measured object. For example, when the object to be measured is the fundus, it is from the boundary between the vitreous body and the retina to the deepest part of the fundus where the probe light can reach. This deepest point is determined by the ratio of signal to noise level.

  Next, in order to separate the region of the retina and choroid (the inner membrane adjacent to the retina), the high light at the boundary between the retinal pigment epithelium (RPE) (the innermost layer of the retina) and the choroidal capillary branch (CC) A reflective layer is extracted.

  In order to remove noise, a Gaussian filter (a filter that performs smoothing based on a Gaussian distribution) having a size of a predetermined pixel (for example, 3 × 3) is applied and averaged. Using this, a differential image of image intensity is obtained. The positive sign portion of the differential image indicates a portion where the image intensity increases, and the negative portion of the differential image indicates a portion where the image intensity decreases. Since the retina has a high light reflectivity and the light intensity is maximized, the portion where the sign of the differential image is positive is the front side of the retina, and the negative portion is the back side of the retina. Thereby, the front side and the rear side of the retina can be detected. A threshold image is applied to the differential image of the intensity, and a binarized boundary image (retinal pigment epithelium) is obtained by connecting positions where the sign changes.

  Although the above description is supplemented here, when the intensity of the image is viewed from the front to the back in the A scan direction, the first boundary surface (the boundary between the vitreous and the retina (retinal pigment epithelium)) is the first reflection. It becomes a surface and is brightly observed. The brightest place can be recognized as a point where the derivative of intensity changes from positive to negative. In addition, the differential is positive (the portion where the intensity increases), and the differential is negative (the portion where the intensity decreases) is the back (rear side). As a result, the first surface (boundary) and its context are automatically obtained.

  In the binarized image, after removing noise components (small volume portions that do not represent the boundary), the pixel intensity is examined in order from the top to the bottom of the three-dimensional image, and is higher than a predetermined threshold value. Find the point where a pixel is first detected. This point is where the increase in image intensity has started, and it can be said that the boundary layer starts. Progressing downward from that point, for example, a point where the intensity change of the image becomes maximum within 10 pixels (a point having the maximum image intensity gradient) is obtained. This point is the front boundary surface of the retina (retinal pigment epithelium).

  A recursive smoothing technique is used to smooth the front boundary. This is a method of determining a boundary surface by linear interpolation after removing a point having a second derivative in the axial direction. Thereafter, for example, a median filter of rank 15 (means for removing noise while preventing blurring of edges in smoothing by giving the median value of pixels in a certain area centered on the pixel instead of the density of a certain pixel) )multiply.

  Next, a highly reflective boundary surface that is a boundary surface between the retina and the choroid is detected. A point having the maximum gradient is detected starting from a point separated from, for example, 25 pixels downward from the front boundary. When no boundary is found in the A scan direction, the A line is ignored. The pixel intensity of the obtained points is averaged, for example, 15 pixels in the optical axis direction. The interface between the retina and choroid is smoothed by filtering if necessary. Here, the “25 pixels” is smaller than the average thickness of the retina, and the boundary between the retina and the choroid should be set to an area farther than 25 pixels from the boundary between the retina and the vitreous body. This is based on prior knowledge.

  A point where the pixel intensity (OCT signal) is extremely small is considered to be a blood vessel region in the retina. This is because of light absorption by blood. In addition, since the light intensity is weakened due to absorption by blood under the blood vessel, the OCT signal becomes small, and the retinal boundary (the boundary between the retina and the choroid) may not be detected. In order to determine the boundary surface, the points where the boundary could not be detected are averaged for, for example, 10 pixels around the point, and the boundary position of the point that could not be detected is complemented using information on the boundary that has already been detected. To do.

  In order to determine the interface between the retina and the choroid, it is necessary to determine the boundary between the signal and the noise magnitude. For that purpose, the intensity image blurred by the Gaussian filter is binarized. If the intensity histogram is taken, the noise has a sharp peak, and for example, 5 levels below the peak is set as a threshold value. This eliminates small noise. The region of the retina is indicated by region D in FIG.

  The boundary between the choroid and the sclera (the lower (back side) surface of the choroid) is also determined in the same manner. The choroid region is indicated by region E in FIG.

  A three-dimensional blood vessel network image is displayed in the retinal volume cut out by the above method. Since blood vessels existing in the retina region have a fast blood flow, the interference fringes become unclear (this phenomenon is called “fringe washout”), and the OCT signal becomes weak. Therefore, by connecting portions where the OCT signal is weak in the retina, a three-dimensional distribution pattern (network) of blood vessels can be displayed separately (see FIG. 6B).

  Although a three-dimensional blood vessel image is displayed in the choroid volume cut out by the above method, Doppler optical coherence tomography can obtain information on the speed (speed and direction) of a moving object. it can. Therefore, the blood flow (blood cell speed) flowing through the blood vessel can be measured in a speed range (when slow) where fringe washout does not occur.

  For blood vessels with slow blood flow, such as choroidal blood vessels, using Doppler information, a blood vessel distribution pattern (network) and a blood vessel network image including the direction of blood flow can be displayed. In order to obtain a clearer image, it may be displayed after taking the square of the speed and eliminating the flow information. The image may be displayed with a three-dimensional Gaussian filter for removing ghost noise and discontinuity (see FIG. 6E).

  By integrating the extracted image in the volume in the direction of the optical axis, a two-dimensional angiographic image can be created. In the case of creating a contrast image, it is possible to obtain a contrast image with better contrast by using the image displayed after taking the square of the speed (without the flow information) (FIGS. 6A and 6D). )reference).

  Here, FIG. 6A is a projection of all measurement region data onto a plane, and corresponds to an image taken by a general fundus camera. FIG. 6A is a projection of only the retina and the choroid, and the blood vessel image can be observed more clearly.

  Since the regions are separated in advance by a highly reflective layer such as the boundary between the vitreous and the retina, the boundary between the retina and the choroid, and the boundary between the choroid and the sclera, separate blood vessels in the retina and blood vessels in the choroid are separated. A contrast image can be created (see FIGS. 6D and 6E).

  FIG. 6 (f) is a composite of the blood vessel images obtained by separation, and it is possible to display more easily, such as displaying the intraretinal blood vessel in green and the choroidal blood vessel in yellow. This is a great advantage of the fourth embodiment, and such an image cannot be obtained by normal fundus angiography.

  The best mode for carrying out the present invention has been described above based on the embodiments. However, the present invention is not limited to such embodiments, and the technical matters described in the claims are not limited. It goes without saying that there are various embodiments within the scope.

  The optical coherence tomography correction method in the optical tomographic image processing method according to the present invention is simple in configuration without using a complicated additional system such as adaptive optics or a motion track, and only the three-dimensional measurement data is used to measure the object to be measured. Since optical distortion can be reduced by correcting the movement, an optical tomographic imaging apparatus in the medical field such as ophthalmology where accurate inspection cannot be performed especially when the image is distorted due to movement or shaking of a biological sample such as fundus examination It is effective when applied as a correction method.

  Further, when the optical tomographic image processing method according to the present invention is used, if there is a different structure in the depth direction of the object to be measured in the distortion-corrected image, the structure and useful information included in the structure are included. Can be automatically separated.

  Furthermore, it becomes easy to extract useful structure information included in the layer structure from the optical tomographic image. This makes it possible to automatically extract information desired for screening such as group health examinations that handle a large amount of data, which is effective for practical application of an optical tomographic imaging apparatus in the medical field such as ophthalmology.

It is a figure which shows the whole structure of FD-OCT which concerns on Example 1 of this invention. FIG. 2 is an explanatory diagram based on a two-dimensional optical tomographic image for explaining the principle and operation of the first embodiment. It is the figure which represented the histogram of phase difference typically. It is a figure which shows the whole structure of PS-FD-OCT which concerns on Example 2 of this invention. It is a figure which shows the whole structure of SS-OCT which concerns on Example 3 of this invention. It is a figure which shows the image of the retina which is Example 4 of this invention, and a choroid. It is a figure explaining the conventional OCT.

Explanation of symbols

1 FD-OCT
2 Broadband light source
3 Low coherence interferometer
4 Spectrometer
5 Beam splitter
6, 9, 23, 32, 34, 35, 36 Lens
7 Galvano mirror
8 Object to be measured
10 Reference mirror
11 Diffraction grating
12 CCD
13 PS-FD-OCT
14 Light wedge
15 Polarizer
16, 22 1/2 wavelength plate
17, 18, 19, 21 1/4 wave plate
20 mirror
24 wavelength scanning OCT
25 wavelength scanning light source
26, 28, 37 fiber
27 Fiber coupler
29, 48 Object to be measured
30 fiber
31 Fixed reference mirror
33 Scanning mirror
38 Light detector
39 computers
40 display
43 OCT
44 Light source
45 Collimating lens
46 Beam splitter
47 Objective lens in the object arm
49 Objective lens in the reference arm
50 reference mirror
51 condenser lens
52 (Photodiode etc.) Photodetector

Claims (6)

Using optical coherence tomography, a plurality of one-dimensional tomographic images are acquired while shifting the position in the direction perpendicular to the axis (A scan axis) in the depth direction (A scan axis) of the object to be measured. A two-dimensional tomographic image (B-scan image) parallel to the axis is acquired, and a plurality of two-dimensional tomographic images are acquired while shifting the position in the vertical direction (C-scan direction) with respect to the two-dimensional tomographic image to form a three-dimensional image In optical tomography by optical coherence tomography,
Detecting misalignment of the plurality of one-dimensional tomographic images due to movement blur of the measured object using Doppler signal information, correcting misalignment of the one-dimensional tomographic image based on the detection result, An optical tomographic image processing method comprising reconstructing a two-dimensional tomographic image and correcting distortion of an image due to movement of an object to be measured in an optical tomographic imaging method of optical coherence tomography. Using optical coherence tomography, a two-dimensional tomographic image (B-scan image) parallel to the axis in the depth direction (A-scan axis) of the object to be measured is positioned in the direction perpendicular to the two-dimensional tomographic image (C-scan direction). In optical tomographic imaging by optical coherence tomography that acquires a plurality of images while shifting and constructs a three-dimensional image,
A positional deviation of each of the plurality of acquired two-dimensional tomographic images caused by the movement blur of the measured object is detected using Doppler signal information, and the positional deviation of the two-dimensional tomographic image is corrected based on the detection result. The optical tomographic image processing according to claim 1, wherein the three-dimensional image is reconstructed and image distortion due to movement of an object to be measured in the optical tomographic imaging method of optical coherence tomography is corrected. Method.   The optical coherence tomography is Doppler optical coherence tomography, and is used to correct a speed measurement error due to movement blur that occurs when obtaining a flow velocity and a movement amount of a measurement target in a tomographic image in the Doppler optical coherence tomography. 3. The optical tomographic image processing method according to claim 1, wherein image distortion caused by movement of an object to be measured in an optical tomographic imaging method of optical coherence tomography is corrected.   The Doppler signal represents the moving speed of the object to be examined and is represented by the amount of change in the phase component of the optical coherence tomography image, and a Doppler signal histogram between A scan data at different positions in the B scan direction is obtained. In the histogram, the most frequent component is regarded as a movement blur, and the distortion of the image due to the movement of the measured object in the optical tomography method of optical coherence tomography is corrected using the value of the component. An optical tomographic image processing method according to claim 1 or 2.   3. The image distortion due to the movement of an object to be measured in an optical tomographic imaging method of optical coherence tomography is corrected using the median value of the Doppler signal as a movement blur. The processing method of the optical tomographic image of description.   6. If there is a different structure in the depth direction of the object to be measured in the image whose distortion has been corrected, the structure and useful information contained in the structure are automatically separated. A method for processing an optical tomographic image according to any one of the above.