JP2016080659A - Light interference tomographic image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a light interference tomographic image processing method suitable for eliminating noises included in a tomographic image acquired by a light interference tomographic imaging method.SOLUTION: A light interference tomographic image processing method according to the present invention is a method for processing a tomographic image acquired by a light interference tomographic imaging method, and includes: (1) a feature amount calculation step of calculating a feature amount in a depth direction of the tomographic image; (2) a position calculation step of calculating a surface position and a depth reach limit position of a measurement object; (3) an estimation step of acquiring an estimation feature amount and an estimation depth reach limit position when brightness is not reduced based on the feature amount and the depth reach limit position; (4) a correction amount calculation step of calculating a correction amount at each position in the depth direction based on the feature amount, the estimation feature amount, the depth reach limit position, the estimation depth reach limit position, and the surface position, and (5) an adding step of correcting the tomographic image by adding the correction amount to intensity distribution in the depth direction of the tomographic image.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a method for processing a tomographic image acquired by optical coherence tomography.

  Optical coherence tomography (OCT) can observe the internal structure of a measurement object with high resolution. OCT has attracted attention in the bio / medical field, and has begun to be used for diagnosis of living bodies as measurement objects. Since the tomographic image acquired by OCT may contain noise, it is desired to make a diagnosis after removing the noise from the tomographic image.

  As a method for removing noise from a tomographic image, methods described in Non-Patent Documents 1 and 2 are known. The noise removal method described in Non-Patent Document 1 removes noise by smoothing a tomographic image with a smoothing filter such as a Gaussian filter. The noise removal method described in Non-Patent Document 2 removes speckle noise by taking an average of a plurality of tomographic images.

J. Rogowska et al., Physics inMedicine and Biology Vol.47, pp.641-655, 2002. M. Szkulmowskiet al., Opt Exp, Vol.20, No.2, pp.1337-1359, 2012.

  According to the knowledge of the present inventor, certain types of noise included in tomographic images acquired by optical coherence tomography can be removed even by using the noise removal methods described in Non-Patent Documents 1 and 2. I can't.

  The present invention has been made to solve the above-described problems, and provides an optical coherence tomographic image processing method suitable for removing noise included in a tomographic image acquired by optical coherence tomography. Objective.

  The optical coherence tomographic image processing method of the present invention is a method for processing a tomographic image acquired by optical coherence tomography, and (1) calculates a feature amount of an intensity distribution in the depth direction of the tomographic image of each line. A feature amount calculating step, (2) a position calculating step for calculating the surface position and depth limit position of the measurement target of each line, and (3) for each line, based on the feature amount and the depth limit position, An estimation step for obtaining an estimated feature amount and an estimated depth limit position when there is no decrease in brightness; and (4) for each line, the feature amount, the estimated feature amount, the depth limit position, the estimated depth limit position, and A correction amount calculating step for calculating a correction amount at each position in the depth direction based on the surface position; and (5) adding the correction amount to the intensity distribution in the depth direction of the tomographic image of each line. Addition step to correct the image And comprising.

  According to the present invention, noise included in a tomographic image acquired by optical coherence tomography can be removed.

It is a figure which shows an example of the tomographic image in which the influence of a motion artifact has appeared. It is a flowchart explaining the optical coherence tomographic image processing method of this embodiment. It is a figure which shows the example of intensity distribution I (i, z) of a one-dimensional tomographic image. It is a figure which approximates the range where intensity | strength is larger than a noise floor among the intensity distribution I (i, z) of a one-dimensional tomographic image by a straight line. It is a figure which shows distribution of correction amount (DELTA) I (I, z). It is a figure which shows intensity distribution I '(i, z) obtained by adding correction amount (DELTA) I (I, z) to intensity distribution I (i, z) of the one-dimensional tomographic image of an artifact part.

  The optical coherence tomographic image processing method of the present invention is a method for processing a tomographic image acquired by optical coherence tomography, and (1) calculates a feature amount of an intensity distribution in the depth direction of the tomographic image of each line. A feature amount calculating step, (2) a position calculating step for calculating the surface position and depth limit position of the measurement target of each line, and (3) for each line, based on the feature amount and the depth limit position, An estimation step for obtaining an estimated feature amount and an estimated depth limit position when there is no decrease in brightness; and (4) for each line, the feature amount, the estimated feature amount, the depth limit position, the estimated depth limit position, and A correction amount calculating step for calculating a correction amount at each position in the depth direction based on the surface position; and (5) adding the correction amount to the intensity distribution in the depth direction of the tomographic image of each line. Addition step to correct the image And comprising.

  In the feature amount calculating step, it is preferable that the integral value of the intensity distribution of the tomographic image in the entire range in the depth direction or a limited range is used as the feature amount, or the entire range in the depth direction or It is also preferable to use the fitting coefficient when the fitting is applied to the intensity distribution of the tomographic image within a limited range as the feature amount.

  The tomographic image to be processed is preferably a tomographic image constructed from a spectrum measured by a spectrometer, and is preferably a tomographic image obtained by rotational scanning of a probe inserted into a tubular measuring object. It is.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same elements are denoted by the same reference numerals, and redundant description is omitted. The present invention is not limited to these exemplifications, but is defined by the scope of the claims, and is intended to include all modifications within the meaning and scope equivalent to the scope of the claims.

  Since OCT is used for diagnosis of living bodies in the bio / medical field, it is important to reduce the burden on the living body (patient) as much as possible. Therefore, OCT is required to shorten the imaging time (that is, to increase the imaging speed).

  OCT is roughly classified into TD-OCT (Time-Domain OCT) and FD-OCT (Fourier-Domain OCT). TD-OCT is a method for acquiring a tomographic image by dynamically changing the optical path length of the reference light in the interferometer during imaging. FD-OCT is a method for obtaining a tomographic image by measuring a spectrum of interference fringes and Fourier-transforming the spectrum. At present, FD-OCT, which does not require a dynamic mechanism, has become the mainstream, in contrast to TD-OCT, which requires a mechanical dynamic mechanism in the interferometer, in accordance with the above-described demand for imaging speed.

  OCT can acquire a one-dimensional tomographic image in the depth direction at a certain position by irradiating measurement light to a certain position of the measurement target. Moreover, OCT can acquire a two-dimensional or three-dimensional tomographic image by scanning the irradiation position of the measurement light with respect to a measuring object. In order to obtain a two-dimensional or three-dimensional tomographic image at a high speed by OCT, it is necessary to accelerate the scanning of the measurement light irradiation on the measurement object.

  However, if the amount of displacement of the measurement light irradiation position by scanning is large with respect to the exposure time of the light receiving element that receives the interference light, the phase of the interference fringe may change during the exposure. In this case, since the intensity of the interference light obtained by the light receiving element is accumulated over the exposure time, the amplitude is attenuated as compared with the stationary time, and the luminance of the tomographic image obtained by Fourier transform is lowered. This phenomenon is called motion artifact.

  SD-OCT (Spectral-Domain OCT), which is a type of FD-OCT, is a technique for acquiring a tomographic image from a spectrum of interference light measured by a spectrometer. Since it is necessary to receive light over a long exposure time, the influence of motion artifact is significant.

  In SD-OCT, assuming that the phase of the interference fringes changes linearly by Δφ within the exposure time, theoretically, the attenuation rate of the amplitude of the interference fringes is expressed by the following equation: sin (Δφ / 2) / (Δφ / 2) And does not depend on the frequency of the interference fringes, that is, the depth in the tomographic image. In addition, since a noise floor exists in an actual tomographic image, attenuation due to the influence of motion artifacts occurs in a range above the noise floor.

  FIG. 1 is a diagram illustrating an example of a tomographic image in which the influence of motion artifact appears. An OCT catheter including an optical fiber having an optical component that reflects measurement light to the side by 90 degrees is attached to the tip, and the OCT catheter is connected to an OCT apparatus. An OCT catheter was placed in an intralipid (artificial fat) solution, and an optical fiber inside the OCT catheter was rotated at high speed to obtain a tomographic image. This figure is a two-dimensional tomographic image formed by arranging a plurality of lines of one-dimensional tomographic images taken during one rotation in a circle. In the circular image shown in the figure, the darkness appears mainly in the directions of 5 o'clock, 8 o'clock and 11 o'clock due to motion artifacts. It is considered that this is because the phase of the interference fringes is fluctuated due to vibrations suddenly generated by high-speed rotation of the optical fiber.

  Such motion artifacts hinder the judgment for the observer and must be removed from the tomographic image. As a method for removing noise from a tomographic image acquired by OCT, removal by a spatial filter described in Non-Patent Document 1 and removal by averaging between frames described in Non-Patent Document 2 are known. However, noise due to this type of motion artifact is not a random pattern, but is attenuated substantially uniformly in the depth direction, and may not be removed by these noise removal methods. In addition, regarding noise removal by averaging between frames, when a moving object is a target, since an image is shifted between frames, an image is blurred by processing. Furthermore, since such processing tends to smooth the tomographic image, it may give a different impression to the observer.

  The optical coherence tomographic image processing method of the present embodiment described below is suitable for removing noise due to motion artifacts from a tomographic image acquired by OCT. The optical coherence tomographic image processing method of the present embodiment is suitable for processing a tomographic image constructed from a spectrum measured by a spectrometer (that is, a tomographic image acquired by SD-OCT), and a tubular measurement. This is suitable for processing a tomographic image obtained by rotational scanning of a probe inserted into a target (for example, a blood vessel).

  FIG. 2 is a flowchart for explaining the optical coherence tomographic image processing method according to this embodiment. The optical coherence tomographic image processing method of this embodiment is a method of processing a tomographic image acquired by OCT, and includes a tomographic image acquisition step S10, a feature amount calculation step S11, a position calculation step S12, an estimation step S13, and a correction amount calculation. Step S14 and addition step S15 are provided.

  In the tomographic image acquisition step S10, a tomographic image to be processed is acquired. This tomographic image is two-dimensional or three-dimensional, and includes a plurality of lines of one-dimensional tomographic images in the depth direction. The intensity distribution of the 1-dimensional tomographic image of the i-th line among these 1-dimensional tomographic images is expressed as I (i, z). z is a variable representing the position in the depth direction.

  In the feature amount calculation step S11, the feature amount S (i) of the intensity distribution I (i, z) in the depth direction of the tomographic image of each line is calculated. This feature amount S (i) represents the degree of luminance reduction due to motion artifact. This feature amount S (i) may be an integral value of the intensity distribution I (i, z) of the tomographic image in the entire range in the depth direction or in a limited range, or the entire amount in the depth direction. It may be a fitting coefficient when fitting is applied to the intensity distribution I (i, z) of the tomographic image in a range or a limited range.

In the position calculation step S12, the surface position z _s (i) and the depth limit position z _e (i) of the measurement target of each line are calculated. The calculation of these positions can be realized by performing edge detection or fitting with a specific function based on the intensity distribution I (i, z) of the tomographic image.

In the estimation step S13, for each line, the estimated feature amount S ′ (i) and the estimated depth limit position z _e ′ (i) when there is no decrease in luminance are used as the surrounding feature amount S (i) and the depth limit. Obtained based on the position z _e (i). Regarding the tendency of the feature quantity S (i) and the depth limit z _e (i) in each line, both the values locally fluctuate in the line where the motion artifact occurs. Therefore, by obtaining the baseline (noise floor), the estimated feature quantity S ′ (i) and the estimated depth limit position z _e ′ (i) when there is no motion artifact can be obtained. The baseline can be obtained, for example, by fitting with a Fourier series, maximum or minimum in the range before and after, or interpolation using a conditional branch.

In the correction amount calculation step S14, for each line, the feature amount S (i), the estimated feature amount S ′ (i), the depth limit position z _e (i), the estimated depth limit position z _e ′ (i), and the surface Based on the position z _s (i), a correction amount ΔI (I, z) at each position in the depth direction is calculated.

  In addition step S15, the tomographic image is corrected by adding a correction amount ΔI (I, z) to the intensity distribution I (I, z) in the depth direction of the tomographic image of each line. Thereby, the intensity distribution I ′ (I, z) (= I (I, z) + ΔI (I, z)) of the tomographic image from which the noise due to the motion artifact is removed can be obtained.

  Next, the optical coherence tomographic image processing method of this embodiment will be further described using the example of the tomographic image of the intralipid solution shown in FIG.

  FIG. 3 is a diagram illustrating an example of the intensity distribution I (i, z) of the one-dimensional tomographic image. The figure shows the intensity distribution of the one-dimensional tomographic image of the line (artifact part) affected by the motion artifact in the tomographic image of the intralipid solution shown in FIG. 1, and the motion located in the vicinity of the artifact part. The intensity distribution of the one-dimensional tomographic image of the line (normal part) without the influence of an artifact is shown.

As shown in the figure, the surface positions z _s (i) of the artifact part and the normal part are the same. In the intensity distribution of any one-dimensional tomographic image of the artifact part and the normal part, the baseline (noise floor) is the same in a range deeper than a certain position in the depth direction. In a range where the intensity is larger than the noise floor, the intensity distribution of the one-dimensional tomographic image of the artifact part is attenuated substantially uniformly as compared with the intensity distribution of the one-dimensional tomographic image of the normal part. In the range where the intensity is greater than the noise floor, the intensity distribution of any one-dimensional tomographic image of the artifact part and the normal part can be linearly approximated with respect to the position variable z in the depth direction.

FIG. 4 is a diagram showing a straight line approximating a range where the intensity is greater than the noise floor in the intensity distribution I (i, z) of the one-dimensional tomographic image. The figure also shows the level of the noise floor, the surface position z _s (i) and the depth limit position z _e (i). The surface position z _s (i) of each of the artifact part and the normal part is the same. However, the depth limit position z _e (i) of the artifact part is shallower than the depth limit position of the normal part. The estimated feature value obtained based on the integrated value S (i) of the intensity distribution of the surrounding one-dimensional tomographic image is S ′ (i), and the integrated value of the intensity distribution of the one-dimensional tomographic image of the artifact portion is S (i ), The difference ΔS (i) (= S ′ (i) −S (i)) is an area of a range surrounded by four solid lines (a range surrounded by a broken line) in FIG. It is. This ΔS (i) is equal to the integral value of the correction amount ΔI (I, z).

  FIG. 5 is a diagram showing the distribution of the correction amount ΔI (I, z). Since the attenuation amount due to the motion artifact may not depend on the position z in the depth direction, the correction amount ΔI (I, z) at each position can be obtained relatively easily by the following equation. The distribution of the correction amount ΔI (I, z) shown in the figure is obtained according to this equation.

  FIG. 6 is a diagram showing an intensity distribution I ′ (i, z) obtained by adding the correction amount ΔI (I, z) to the intensity distribution I (i, z) of the one-dimensional tomographic image of the artifact portion. . As shown in the figure, the intensity distribution I ′ (i, z) of the one-dimensional tomographic image of the artifact part corrected in this way has the same tendency as the intensity distribution of the one-dimensional tomographic image of the normal part. It has become.

  As described above, the optical coherence tomographic image processing method of the present embodiment only gives an offset to the original tomographic image, and is not a process of directly changing the distribution like a spatial filter or a frame average. Noise due to motion artifacts can be effectively removed without changing the image quality.

Claims (5)

A method for processing a tomographic image acquired by optical coherence tomography,
A feature amount calculating step for calculating a feature amount of the intensity distribution in the depth direction of the tomographic image of each line;
A position calculating step for calculating the surface position and depth limit position of the measurement target for each line;
For each line, based on the feature amount and the depth limit position, an estimation step for obtaining an estimated feature amount and an estimated depth limit position when there is no decrease in brightness; and
A correction amount calculating step for calculating a correction amount at each position in the depth direction based on the feature amount, the estimated feature amount, the depth limit position, the estimated depth limit position, and the surface position for each line. When,
An addition step of correcting the tomographic image by adding the correction amount to the intensity distribution in the depth direction of the tomographic image of each line;
An optical coherence tomographic image processing method. In the feature amount calculation step, an integral value of the intensity distribution of the tomographic image in the entire range in the depth direction or a limited range is used as the feature amount.
The optical coherence tomographic image processing method according to claim 1. In the feature amount calculating step, a fitting coefficient when fitting is applied to the intensity distribution of the tomographic image in the entire range or a limited range in the depth direction is set as the feature amount.
The optical coherence tomographic image processing method according to claim 1. The tomographic image to be processed is a tomographic image constructed from the spectrum measured by the spectrometer.
The optical coherence tomographic image processing method according to claim 1. The tomographic image to be processed is a tomographic image obtained by rotational scanning of a probe inserted into a tubular measurement target.
The optical coherence tomographic image processing method according to claim 1.