JP2016003898A - Optical tomographic imaging apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an optical tomographic imaging apparatus capable of acquiring a two-dimensional tomographic image and/or a three-dimensional tomographic image having higher SNR and image quality by spatially removing noise of a high frequency and executing averaging processing more effectively in the optical tomographic imaging apparatus for acquiring the two-dimensional tomographic image and/or the three-dimensional tomographic image corresponding to a Jones matrix acquired by a polarization-sensitive OCT.SOLUTION: An optical tomographic imaging apparatus includes processing means for averaging and processing each pixel in a prescribed kernel by setting the prescribed kernel (prescribed pixel range in an XZ direction and/or an XYZ direction) to a two-dimensional tomographic image and/or a three-dimensional tomographic image corresponding to a Jones matrix acquired by a polarization-sensitive OCT and by executing a given weighing processing to each pixel in an X direction and/or a Y direction and/or a Z direction of the set prescribed kernel.

Description

The present invention relates to an optical tomographic imaging apparatus, and more particularly to an optical tomographic imaging apparatus that takes a tomographic image by OCT used in ophthalmic medicine or the like.

Optical coherence tomography (OCT) is a method widely used as a means for acquiring high-resolution tomographic images of living tissue in ophthalmology because it can be measured non-invasively and non-contactly.

In optical coherence tomography (OCT), a time domain method called a time domain method, a time domain OCT that acquires a tomographic image while mechanically changing the optical path length of reference light by moving a mirror, and a spectroscope called a Fourier domain method are used. There are spectral domain OCT that uses spectral information to detect tomographic images, and optical frequency swept OCT that detects spectral interference signals using a wavelength scanning light source to acquire tomographic images.

Birefringence that changes the polarization state occurs in tissues in which molecules are arranged in a certain direction. In the retina at the fundus, there is strong birefringence in the retinal nerve fiber layer, retinal pigment epithelium layer, blood vessel wall, sclera, and phloem. In recent years, the development of various polarization-sensitive OCTs has been attempted for polarization-sensitive OCT (PS-OCT), which is one of functional OCTs, in order to visualize these tissues by forming a birefringent tomography.

Polarization-sensitive OCT (PS-OCT) employs a configuration in which circularly polarized light or polarization-modulated light is used as measurement light for observing a sample, and interference light is detected as two orthogonal linearly polarized lights.

Patent Document 1 discloses an example of polarization-sensitive OCT (PS-OCT). At the same time (synchronously) with the B-scan, a polarized beam from a light source (a beam linearly polarized by a polarizer) is continuously modulated by an EO modulator (polarization modulator, electro-optic modulator). Then, the polarized light beam whose polarization is continuously modulated is divided, one of the samples is irradiated, and the reflected light is obtained, and the other is used as the reference light, and OCT measurement is performed by spectral interference between the two. Among them, it is disclosed that a Joan matrix representing a polarization characteristic of a sample is obtained by simultaneously measuring a vertical polarization component and a horizontal polarization component with two photodetectors.

In Non-Patent Document 1, in order to improve the SNR (signal-to-noise ratio) and image quality of an image corresponding to the Jones matrix acquired by polarization-sensitive OCT (PS-OCT), the B-scan image corresponding to the Jones matrix is For example, a method of moving average with a kernel size of 3 × 5 pixels is disclosed.

Japanese Patent No. 434429

"Birefringence measurement of conena and anientor segment by office-basic polarization-sensitive optical coherence tomography," BIOMEGITAL OPTICS. 2, no. 8, August 1, 2011, 2392

However, the moving average disclosed in Non-Patent Document 1 simply averages each Jones matrix element of each pixel of the kernel. In this case, averaging is performed after the rectangular wave processing of 0/1 is performed in the X direction and Z direction of the kernel.

Square wave processing in real space is equivalent to multiplying the SINC function in Fourier space. FIG. 5 shows the spatial frequency characteristics of the SINC function (solid line) and the Gaussian function (dotted line) in Fourier space. As described above, since the rectangular wave processing in the real space is equivalent to the multiplication of the SINC function in the Fourier space, when the rectangular wave processing is performed as in Non-Patent Document 1, as shown by the solid line in FIG. There is a problem that high-frequency noise (for example, speckle noise) remains without being cut (removed), and as a result, a high-quality image cannot be acquired.

The present invention solves the above-described problem and provides an optical tomographic imaging apparatus that acquires a two-dimensional and / or three-dimensional tomographic image corresponding to a Jones matrix acquired by polarization-sensitive OCT (PS-OCT). An optical tomographic imaging apparatus capable of acquiring two-dimensional and / or three-dimensional tomographic images having higher SNR and image quality by spatially removing high-frequency noise and performing averaging more effectively The purpose is to provide.

In order to achieve the above object, the present invention relates to a polarization-sensitive OCT (PS-OCT) that acquires a Joan matrix that represents the polarization characteristics of a test object, and a B-scan image acquired corresponding to the Jones matrix or A predetermined kernel (a predetermined pixel range in the XZ direction or / or XYZ direction) is set for the / and C scan image (volume data), and the set X direction or / and Y direction or / and Z of the predetermined kernel A processing means is provided for performing a predetermined weighting process on each element of the Jones matrix of each pixel in the direction and averaging each element of the Jones matrix of each pixel in a predetermined kernel.

In the present invention, as described above, the weighting process is performed on each element of the Jones matrix of each pixel of the kernel. For example, when the kernel is two-dimensional, Gaussian waveform processing of 0 to 1 is performed in the X direction and the Z direction and averaged.

As described above, performing Gaussian waveform processing in real space is equivalent to multiplying the Gaussian function (dotted line) in FIG. 5 (in the Fourier-transformed measurement signal) in Fourier space. It can be seen that the noise can be removed efficiently. Furthermore, more effective averaging processing can be performed by performing weighting processing.

According to the present invention, by performing weighting processing such as Gaussian waveform processing, spatially high frequency noise such as speckle noise can be removed and averaging processing can be effectively performed. A two-dimensional and / or three-dimensional tomographic image having image quality can be acquired.

It is the figure which showed the detail of an example of the optical tomographic image acquisition part by this invention. It is the figure which showed the structure of the optical tomography apparatus. It is the figure which showed the flowchart of a moving average process. It is a figure explaining the flow until acquisition of a three-dimensional tomogram. It is the figure which showed the spatial frequency characteristic of the SINC function (solid line) and the Gaussian function (dotted line) in Fourier space.

An optical tomographic imaging apparatus according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a detailed configuration of the tomographic image acquisition unit 100.

As shown in FIG. 1, the tomographic image acquisition unit 100 shoots a two-dimensional or / and three-dimensional tomographic image of the sample 117 by irradiating the sample (inspection object) 117 with measurement light. In this embodiment, a Fourier domain (optical frequency sweep) method using a wavelength scanning light source 101 that scans while changing the wavelength with time is employed.

That is, the light output from the wavelength scanning light source 101 is input to the fiber coupler 102 through the optical fiber, and is split into the reference light and the measurement light at the ratio of, for example, 5:95, and is respectively referred to. It is output to the arm 160 and the sample arm 150. Of these, the reference light output to the reference arm 160 is input to the optical circulator 120 through the optical fiber, is then input to the collimator lens 121, and is incident on the reference mirror 122. The reference mirror 122 can be moved and controlled on the optical path axis to adjust the optical path length for aligning the reference optical path with the surface position of the sample, and the measurement optical path length and the reference optical path length are matched before measuring the OCT tomographic image.

Then, the reference light reflected by the reference mirror 122 passes through the optical fiber from the collimator lens 121, the optical path is changed by the optical circulator 120, passes through the polarization controller 119, enters the collimator lens 123, and enters the polarization sensitive detection arm 136. The

On the other hand, the measurement light output from the fiber coupler 102 to the sample arm 150 is input to the collimator lens 104 of the polarization dependent delay line 133 through the optical fiber via the polarization controller 103 and then passes through the polarizer 105. In this embodiment, the polarization angle of the polarizer 105 is set to 45 degrees. Furthermore, the polarization angle immediately before passing through the polarization controller 103 and entering the collimator lens 104 is also controlled to 45 degrees, and the polarization controller 103 and the polarizer 105 are adjusted and adjusted so that measurement light polarized at 45 degrees can be efficiently extracted. It is controlled.

The measurement light polarized at 45 degrees passes through the polarization beam splitter 106 in the polarization-dependent delay line 133 and is split into light in two linear polarization states (vertical direction and horizontal direction) orthogonal to each other. The divided measurement lights are reflected by different total reflection prisms 107 and 108, respectively, and propagate in two different optical paths. Here, the movement between at least one of the total reflection prisms 107 and 108 is controlled to cause a delay between two different polarization states (vertical direction and horizontal direction).

Here, by setting the incident measurement light so as to be incident on a position deviated from the center of the polarizing beam splitter 106, light having two different polarization states is generated by the polarizing beam splitter 106, and each has a different total reflection prism. Measurement light reflected in 107 and 108 and having two different polarization states (vertical direction and horizontal direction) having a certain delay is generated. The optical path is changed by the reflection mirror 110, and then the optical fiber is connected by the collimator lens 109. .

After the measurement light passing through the optical fiber passes through the polarization controller 111, the optical path is changed by the optical circulator 112, enters the collimator lens 113, is reflected by the galvanometer mirrors 114 and 115, is condensed by the lens 116, It enters the sample (inspection object) 117.

The galvanometer mirrors 114 and 115 are for scanning the measurement light. By controlling the galvanometer mirrors 114 and 115, the measurement light is scanned in the horizontal direction and the vertical direction on the surface of the sample 117. Yes. Thereby, a two-dimensional tomographic image and a three-dimensional tomographic image of the sample 117 can be acquired.

The measurement light reflected by the sample (inspection object) 117 passes through the lens 116 and the galvanometer mirrors 115 and 114, and is input to the collimator lens 113. Then, the optical path of the measurement light is changed by the optical circulator 112 through the optical fiber, and after passing through the polarization controller 118, is input to the collimator lens 125 and input to the polarization sensitive detection arm 136.

The reference light that is input from the collimator lens 123 to the polarization sensitive detection arm 136 and is polarized by the polarizer 124 and the measurement light that is reflected by the sample (inspection object) 117 are combined using a non-polarizing beam splitter 132, Divided. The split light is then input to collimator lenses 126 and 127 and divided into two orthogonal polarization states by two in-line polarization beam splitters 128 and 129.

Here, in order to equalize the power of the linearly polarized light in the vertical and horizontal directions of the reference light after the inline polarization beam splitters 128 and 129, the polarization angle of the polarizer 124 is adjusted to 45 degrees and the efficiency is increased. Therefore, the polarization controller 119 passing in advance is controlled so that the polarization angle immediately before entering the polarizer 124 is approximately 45 degrees.

The interference between the two polarization states is detected by the two balanced photodetectors 130 and 131. The detected interference signals of two polarization states in the vertical direction and the horizontal direction are subjected to processing such as Fourier transform on each interference signal in the arithmetic unit 202 provided in the control device 200 shown in FIG. Object) A B-scan image and / or a C-scan image (volume data) acquired corresponding to the Jones matrix of 117 is acquired. The acquired tomographic image is stored in the storage unit 203.

FIG. 4 shows a state in which a tomographic image (B-scan image) is acquired by the tomographic image acquisition unit 100. 4A shows an example of the fundus retina of the eye E, and FIG. 4B shows a plurality of two-dimensional tomograms (B-scan images) of the fundus retina 401 obtained from the tomogram acquisition unit 100. An example is shown. FIG. 4C shows an example of a C-scan image (also referred to as a three-dimensional tomographic image or volume data) of the fundus generated in the present embodiment. 4A to 4C, the x-axis indicates the B-scan scan direction, and the y-axis indicates the C-scan direction. Furthermore, the z-axis in FIGS. 4B and 4C indicates the depth direction of the A scan signal, that is, the depth direction of the fundus.

Reference numeral 404 in FIG. 4B is an acquired two-dimensional tomographic image, which is created by the calculation unit 202 reconstructing the A scan signal 403 while scanning the galvanometer mirror unit 106 in the X direction. This two-dimensional tomographic image is a B-scan image, and is a two-dimensional cross section in the X direction perpendicular to the depth direction (Z direction) with respect to the fundus retina 401, that is, a plane defined by the x axis and the z axis in FIG. It is a two-dimensional tomogram in FIG. Reference numeral 402 in FIG. 4A indicates the photographing position of the two-dimensional tomographic image 404.

The B scan image and the C scan image shown in FIG. 4 are images composed of a plurality of pixels. The moving average according to the present invention is a predetermined number of pixels, for example, in a B-scan image (two-dimensional image), an area having a predetermined number of pixels with respect to the X direction and the Z direction (for example, 3 pixels in the X direction and 5 pixels in the Z direction). If there is, an area of 3 × 5 pixels) is defined as a predetermined kernel area in the present invention. When moving average processing is performed on a C-scan image, a predetermined number of pixel areas in the X, Y, and Z directions are set as predetermined kernel areas.

Next, the moving average processing of the acquired image, which is the point of the present invention, will be described using the flowchart of FIG.

The acquired image is subjected to moving average processing in order to improve SNR (signal noise ratio) and image quality. The flow shown in FIG. 3 is an example of the moving average process, and the moving average process is not limited to this flow.

First, in S10, the first pixel is set. Since the first pixel is the 0th pixel, n = 0 is set in S10, and the set pixel is the nth pixel.

A predetermined kernel area is extracted from the volume data stored in the storage unit 203 in S12 based on the nth pixel. For example, when moving average processing is performed on one B-scan image, a region of a predetermined number of pixels in the X and Z directions including the nth pixel (in the present invention, described as “kernel region”) E.g. 3 pixels x 5 pixels). The relationship between the kernel area and the position of the nth pixel may be set in advance. If the relationship is constant, the extraction is easy and the effect of the moving averaging process can be easily obtained.

In addition, when moving average processing is performed on volume data, an area of a predetermined number of pixels in each of the X direction, the Y direction, and the Z direction, including the nth pixel (eg, 3 × 5 × 3). Pixel area).

In S14, a window function is applied to the extracted kernel area. As the direction of multiplication, if the target is a B-scan image, the direction may be the X and Z directions, and if the target is volume data, it may be multiplied in the X, Y, and Z directions, and the effect can be expected in any one direction. In this case, any one of the X, Y, and Z directions may be used. In the case of volume data, two selected directions may be used.

As a window function, a Gaussian function (also referred to as a Gaussian window) may be used. When the Gaussian function is used, high-frequency noise can be efficiently removed from the spatial frequency characteristic indicated by the dotted line in FIG.

The process of S14 multiplies each element of the Jones matrix of each pixel by a window function such as a Gaussian function. The window function to be multiplied is not limited to the Gaussian function, and a Hamming function (also referred to as a Hamming window) or a Hanning function (also referred to as a Hanning window) may be employed. Even if these functions are employed, the same effect as in the case of the Gaussian function can be obtained.

After the window function is applied to the kernel area in S14, the SNR (signal-to-noise ratio) of each pixel in the kernel area is calculated in S16, and the pixel having the highest SNR is detected. .

In S18, a global phase difference between each Jones matrix element of the pixel having the highest SNR detected in S16 and each Jones matrix element of the other pixels is calculated. The global phase difference is calculated for each element of the Jones matrix of each pixel.

In S20, the global phase difference of each Jones matrix element of each pixel calculated in S18 is canceled using the global phase difference of each Jones matrix element of each pixel. This cancels the global phase difference of each pixel in the kernel region.

After canceling the global phase difference of each pixel in the kernel area in S20, the element values are averaged for each element of the Jones matrix of each pixel in the kernel area in S22.

When the Jones matrix elements are averaged in S22, one pixel number is added in S24. If the added value of n is larger than a predetermined number n (final) in S26, it is determined that the moving average process has been completed in the entire range of the target B-scan image or volume data, and the moving average process ends. To do. If the number is less than the predetermined number n (final), the process returns to S12, the kernel area for the next pixel is extracted, and the processes of S12 to S24 are repeated.

The predetermined number n (final) in S26 may be the total number of pixels of the target B-scan image or volume data. In this case, moving average processing is performed over the entire area of the target B-scan image or volume data. Not limited to this, a predetermined number smaller than the total number may be used. For example, if it is desired to confirm an image of only an arbitrary feature portion in the B-scan image or volume data, the value of n may be set according to the region of the arbitrary feature portion.

As described above, weighted processing such as Gaussian waveform processing is performed and moving average processing is performed, so that spatially high frequency noise such as speckle noise can be removed and averaging processing can be performed effectively. Thus, a two-dimensional and / or three-dimensional tomographic image having a higher SNR and image quality can be acquired.

The embodiments of the present invention have been described in detail above. However, these are merely examples, and the present invention is not construed as being limited by specific descriptions in the embodiments. The present invention can be carried out in a mode in which various changes, modifications, improvements, etc. are added based on the knowledge, and such a mode is within the scope of the present invention as long as it does not depart from the gist of the present invention. It should be understood that it is included in.

100 .. Tomographic image acquisition unit 101. Wavelength scanning light source 102. Fiber coupler 106. Polarization beam splitter 112, 120. Optical circulators 128, 129. Inline polarization beam splitter 130, 131. Balanced light detection. ... Non-polarizing beam splitter 133.. Polarization dependent delay line 201.. AD board 202.

Claims (3)

A polarization-sensitive OCT (PS-OCT) for obtaining a Joan matrix representing the polarization characteristics of a test object,
A predetermined kernel (a predetermined pixel range in the XZ direction or / or XYZ direction) is set for the B scan image or / and C scan image (volume data) acquired corresponding to the Jones matrix,
A predetermined weighting process is performed on each element (pixel value) of the Jones matrix of each pixel in the X direction or / and Y direction or / and Z direction of the predetermined kernel set,
An optical tomographic imaging apparatus comprising processing means for averaging each element (pixel value) of a Jones matrix in a predetermined kernel. The optical tomography apparatus according to claim 1, wherein the predetermined weighting process is a Gaussian process. 2. The optical tomography apparatus according to claim 1, wherein the predetermined weighting process is a window function process such as Hamming or Hanning.