KR102042342B1 - Method for detecting position of radiation source using rotating modulation collimator and imaging device for performing the method - Google Patents

Abstract

In the radiation source position detection method using the rotation modulation collimator according to the present invention, by performing a dispersion stabilization transformation of the modulation pattern according to the number of photons detected by the rotation of the rotation modulation collimator (RMC) and the average of the modulation pattern and Calculating a first detection coefficient from which the correlation of variance has been removed; Calculating a second detection coefficient from which noise is removed by applying a weight calculated based on a similarity between a patch centered on a first angle and a second angle of the rotation modulation collimator to the first detection coefficient; And reconstructing the image by estimating the distribution of the radiation source based on a probability that the calculated radiation source is detected according to the second detection coefficient and the rotation angle of the rotation modulation collimator.

Description

FIELD OF DETECTION OF RADIATION SOURCE USING ROTARY-modulated collimator and imaging device for performing the same

The present invention relates to a method for detecting a radiation source position using a rotation modulated collimator and an imaging apparatus for performing the same.

The Chernobyl nuclear explosion in the former Soviet Union caused severe environmental leaks and caused environmental disasters. In the Fukushima nuclear accident in Japan, high levels of radioactivity leak into the surrounding atmosphere and ocean, which is still difficult to deal with.

As such, if an accident occurs in a nuclear power plant or a radiation-related facility, early blockage of radioactive leakage is treated as the most urgent and important response procedure to prevent large accidents or environmental disasters. For early response and treatment of radiation leakage incidents, radiation detection and radioactive contamination imaging techniques are needed to quickly identify and locate radiation sources.

In this regard, the radiographic imaging apparatus minimizes the possibility of inhalation of the leaked radiation pollutant in the body to detect the location of the radiation source while ensuring the safety of the operator, and uses it to quickly establish a plan for the detection and removal of the radiation pollutant. .

Korean Patent Publication No. 10-2017-0000652

An object of the present invention is to provide a radiation source position detection method using a rotation modulation collimator for calculating a result of removing the noise of the modulation pattern.

It is another object of the present invention to provide an imaging apparatus that accurately detects the position of a radiation source even when the number of detected photons is small.

According to an embodiment of the present invention, a method for detecting a radiation source position using a rotation modulated collimator may perform dispersion stabilization transformation of a modulation pattern according to the number of photons detected by the rotation of a rotation modulation collimator (RMC). Calculating a first detection coefficient from which the association between the mean and the variance has been removed; Calculating a second detection coefficient from which noise is removed by applying a weight calculated based on a similarity between a patch centered on a first angle and a second angle of the rotation modulation collimator to the first detection coefficient; And reconstructing the image by estimating the distribution of the radiation source based on a probability that the calculated radiation source is detected according to the second detection coefficient and the rotation angle of the rotation modulation collimator.

In addition, the method for detecting the position of the radiation source using the rotation modulation collimator according to the embodiment of the present invention may further include calculating a weight by applying a predetermined filter when the number of detected photons is equal to or less than a threshold.

The reconstructing of the image may induce a repetition algorithm by applying a maximum likelihood expectation maximization algorithm (MLEM) based on the distribution of the second detection coefficients.

In addition, the imaging apparatus according to the embodiment of the present invention, the first detection is removed from the correlation between the average and the variance of the modulation pattern by performing a dispersion stabilization transformation on the modulation pattern according to the number of photons detected by the rotation of the rotation modulation collimator A conversion unit for calculating a coefficient; Calculating a second detection coefficient from which noise is removed by applying a weight calculated based on a similarity between a patch centered on a first angle and a second angle of the rotation modulation collimator to the first detection coefficient calculated by the converter; Filtering unit; And a detector for estimating the distribution of the radiation source based on a probability that the calculated radiation source is detected according to the second detection coefficient calculated by the filtering unit and the rotation angle of the rotation modulation collimator. Include.

The imaging apparatus according to an embodiment of the present invention may further include an output unit configured to reconstruct an image of the radiation source based on the position information acquired by the detector.

Radiation source position detection method using a rotationally modulated collimator according to an embodiment of the present invention and an imaging apparatus for performing the same are distributed stabilization transformation and weighting in an environment where the radiation emitted from the radiation source is weak or the number of photons detected by shielding the radiation source is small Iterative algorithm derived by applying noise filtering and Maximum Likelihood Expectation Maximization (MLEM) algorithm to minimize the modulation pattern obtained from the rotation modulation collimator and the noise generated during the imaging process, which improves reliability The detection result can be provided.

1 is a flow chart of a radiation source position detection method using a rotation modulated collimator according to an embodiment of the present invention.
2 is a schematic diagram of an imaging apparatus according to another exemplary embodiment.
3 is a schematic diagram for explaining a mask of the rotation modulation collimator used in the experimental example of the present invention.
4 shows a modulation pattern of photons detected using a rotation modulation collimator, which is a control of the present invention.
5 illustrates a first modulation pattern of photons obtained through the dispersion stabilization transformation step according to an embodiment of the present invention.
6 shows a modulation pattern of photons obtained by applying a Gaussian filter after the dispersion stabilization transformation step, which is a control of the present invention.
FIG. 7 illustrates a second modulation pattern of photons obtained through a radiation source position detection method using a rotation modulation collimator according to an embodiment of the present invention.
8 is a graph illustrating a method of detecting a radiation source position using a rotation modulation collimator and an imaging result of each control group according to an exemplary embodiment of the present invention.
9 is a graph showing a result of performing peak signal to noise ratio (PSNR) in radiation sources of various intensities.

Hereinafter, with reference to the contents described in the accompanying drawings will be described in detail the present invention. However, the present invention is not limited or limited by the exemplary embodiments. Like reference numerals in the drawings denote members that perform substantially the same function.

The objects and effects of the present invention may be naturally understood or more apparent from the following description, and the objects and effects of the present invention are not limited only by the following description. In addition, in describing the present invention, when it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

1 is a flowchart of a method for detecting a radiation source position using a Rotating Modulation Collimator (RMC) according to an embodiment of the present invention.

The radiation source position detection method using the rotation modulation collimator according to the present invention is to improve the problem that the conventional radiation imaging apparatus and the radiation detection method using the same does not detect the exact position due to noise when detecting the position of the radiation source with a small number of photons. .

To this end, the present invention performs a dispersion stabilization transformation on the modulation pattern obtained by the rotation of the rotation modulation collimator to remove the correlation between the mean and the dispersion, and the weight calculated according to the similarity of the patch centered on the angle of the rotation modulation collimator By removing the noise, the location of the radiation source with improved reliability can be estimated even when the radiation emitted from the radiation source is weak or the number of photons detected by the obstacle is insignificant.

Referring to FIG. 1, a radiation source position detection method using a rotation modulation collimator according to the present invention includes a dispersion stabilization conversion step S1, a noise removal step S3, and an image reconstruction step S5.

In the dispersion stabilization transformation step (S1), the dispersion stabilization transformation may be performed on a modulation pattern according to the number of photons detected by the rotation of the rotation modulation collimator to calculate a first detection coefficient from which the correlation between the average of the modulation patterns and the dispersion is removed. .

In the noise removing step S3, the second detection coefficient from which the noise is removed may be calculated by applying a weight calculated based on the similarity of the patch centering on the first angle and the second angle of the rotation modulation collimator to the first detection coefficient. have.

The image reconstruction step S5 may reconstruct the image by estimating the distribution of the radiation source based on the probability that the calculated radiation source is detected according to the second detection coefficient and the rotation angle of the rotation modulation collimator. The image reconstruction step S5 may derive an iterative algorithm by applying a maximum likelihood expectation maximization algorithm (MLEM) based on the distribution of the second detection coefficients.

In addition, the method for detecting the position of the radiation source using the rotation modulation collimator may further include calculating a weight by applying a predetermined filter when the number of detected photons is equal to or less than a threshold.

On the other hand, the radiation source position detection method using a rotation modulation collimator may be performed through the imaging device (1, 2) described later. An imaging algorithm may be required to perform the radiation source position detection method using the rotation modulation collimator. Hereinafter, the description of the imaging apparatus 1 (FIG. 2) and the radiation source position detection method using the rotation modulation collimator will be described in detail.

Meanwhile, the radiation source position detection method using the rotation modulation collimator performed through the above-described steps may be implemented in an application form or in the form of program instructions that can be executed through various computer components and recorded in a computer-readable recording medium. . The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination.

The program instructions recorded on the computer-readable recording medium are those specially designed and configured for the present invention, and may be known and available to those skilled in the computer software arts.

Examples of computer readable recording media include magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD-ROMs, DVDs, and magneto-optical media such as floptical disks. media), and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like.

Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the process according to the invention, and vice versa.

2 is a schematic diagram of an imaging device 1 according to another embodiment of the present invention.

Referring to FIG. 2, the imaging apparatus 1 includes a converter 11, a filter 13, a detector 15, and an output 17. Each component of the imaging device 1 may be formed of an integrated module or of one or more modules. Alternatively, on the contrary, each component may be formed as a separate module.

Since the imaging device 1 has a relatively simple structure, it is easy to carry and can be used even in a narrow space. In addition, the imaging device 1 may be connected to a Rotating Modulation Collimator (RMC) or a communication device to remotely detect the location of the radiation source, but the connection or use of the imaging device 1 is not limited thereto. Do not. In addition, the imaging device 1 may be provided by adding or removing components for performing the radiation source position detection method using the above-described rotation modulation collimator.

Through the rotation modulation collimator 9, a modulation pattern of photons emitted from the radiation source can be obtained. The obtained photon modulation pattern can be modeled as a Poisson distribution. The rotation modulated collimator 9 may comprise two parallel masks with a plurality of slits and a detector. As the rotation modulation collimator 9 rotates, the area of the open space varies at each angle, and a modulation pattern can be generated based on the number of photons counted by the detector. From this modulation pattern, the location of the radiation source can be visualized.

The transform unit 11 may perform a dispersion stabilization transformation on a modulation pattern according to the number of photons detected by the rotation of the rotation modulation collimator 9 to calculate a first detection coefficient from which the correlation between the mean and the variance of the modulation patterns is removed. have. For example, the transformer 11 may perform dispersion stabilization transformation on the Poisson distribution of the modulation pattern acquired by the rotation modulation collimator 9 to calculate the first detection coefficient.

The converter 11 may remove noise from which the degree varies depending on the angle from the observation value of the modulation pattern obtained by the Poisson distribution from the rotation modulation collimator 9. The first detection coefficient calculated by the transform unit 11 refers to each pattern in which the dispersion stabilization transformation is performed on the modulation pattern acquired by the rotation modulation collimator 9, and the first detection coefficient means a set of such first detection coefficients. The modulation pattern can be approximated with a Gaussian distribution with a variance of 1.

The distributed stabilization transform performed by the transform unit 11 may include a Bartlett transform, but is not limited thereto. Detailed features regarding the dispersion stabilization transformation and the Bartlett transformation will be described in detail in the radiation source position detection method.

The filtering unit 13 applies a weight calculated based on the similarity of the patch centering on the first angle and the second angle of the rotation modulation collimator 9 to the first detection coefficient calculated by the transform unit 11 to reduce noise. The removed second detection coefficient can be calculated. The filtering unit 13 calculates the second detection coefficient by performing a non-local means (NLM) algorithm for removing the noise included in the modulated pattern through which the modulated pattern passes. can do.

In particular, the filtering unit 13 is configured to remove noise of the modulation pattern obtained by the rotation modulation collimator 9, and at an angle showing a similar structure based on a patch instead of using a peripheral value of a specific angle when filtering. Can be used, and in detail, the second detection coefficient from which the noise is removed can be calculated by applying a non-local averaging algorithm directly to each of the first detection coefficients on which the transforming unit 11 performs the dispersion stabilization transformation. .

)를 계산하는 과정에서 평균 필터(mean filter)를 선택적으로 적용할 수 있다. 비지역 평균 알고리즘에 관한 자세한 특징은 방사선원 위치 검출 방법에서 상술하도록 한다.In the present invention, a result of applying the non-local averaging algorithm to the first detection coefficients is defined as a second detection coefficient, and the second modulation pattern, which is a set of the second detection coefficients, may generally exhibit a Gaussian distribution. In addition, if the radiation emitted from the radiation source is weak or if the number of detected photons is small, the weight (

You can optionally apply a mean filter in the process of calculating. Detailed features of the non-regional average algorithm will be described later in the radiation source position detection method.

The detector 15 estimates the distribution of the radiation source based on the probability that the calculated radiation source is detected according to the second detection coefficient calculated by the filtering unit 13 and the rotation angle of the rotation modulation collimator 9, and thus the position information of the radiation source. Can be obtained. The detector 15 may acquire location information of the radiation source by applying a maximum likelihood expectation maximization (MLEM) algorithm to the pattern passing through the filter 13.

The detector 15 may apply a maximum likelihood expected value maximization algorithm to the second modulation pattern. Through this, the detector 15 may acquire location information of the radiation source by obtaining and performing an iterative algorithm. Detailed features regarding the maximum likelihood expectation maximization algorithm are detailed in the radiation source position detection method.

The output unit 17 may reconstruct an image of the radiation source position based on the position information acquired by the detector 15. The output unit 17 may output the position of the radiation source based on the coordinate system. The coordinate system used may include, but is not limited to, a one-dimensional coordinate system, a two-dimensional polar coordinate system, a two-dimensional rectangular coordinate system, a three-dimensional cylindrical coordinate system, a three-dimensional spherical coordinate system, and a three-dimensional rectangular coordinate system according to the type of the output unit 17. Do not.

Hereinafter, according to the embodiment of the present invention The radiation source position detection method using the rotation modulation collimator 9 will be described in detail. In the following description, overlapping portions related to the configuration and characteristics of the imaging apparatus 1 will be omitted, and a description related to the radiation source position detection method using the rotation modulation collimator 9 will be described.

)은 회전 변조 시준기(9)를 통해 획득한 변조 패턴일 수 있으며, 변조 패턴(

)은 프아송 분포를 나타낼 수 있다. 회전 변조 시준기(9)는 방사선원의 위치를 3차원 직교좌표계를 기초로 인식할 수 있다. 획득한 변조 패턴(

)을 기초로 한 프아송 분포는 하기의 [수학식 1]과 같이 모델링 될 수 있다.Initially, the modulation pattern provided to the imaging device 1

) May be a modulation pattern obtained through the rotation modulation collimator 9, and the modulation pattern (

) May represent a Poisson distribution. The rotation modulation collimator 9 may recognize the position of the radiation source based on the three-dimensional rectangular coordinate system. Acquired modulation pattern (

The Poisson distribution based on) can be modeled as in Equation 1 below.

[Equation 1]

는 각도 i에서 머무는 시간,

는 방사선원에서 방출되는 방사능,

는 에너지에 따른 회전 변조 시준기(9) 검출기의 검출효율이다.

는 입체각으로 3차원 직교좌표계의 임의의 일 지점인 (x, y, z)에 위치한 방사선원에서 방출된 방사선이 마스크의 앞면으로 들어올 확률이고, 전면에 위치한 마스크로 들어온 방사선이 열린 공간을 통해 검출기까지 도달할 확률을

로 정의하고, 마스크에서 방사선이 차단될 확률을

로 정의하며 두 확률의 합은 1이다. 이때, 마스크의 열린 공간이 아닌 곳에서 광자가 도달하더라도 물질과 에너지에 따라 방사선이 투과하는 비율은

이며,

는 백그라운드 입자 수를 의미한다. In this case, each character used may have the same meaning in the following. In detail, i is the rotation angle of the rotation modulation collimator 9 mask,

Dwell time at angle i,

Is the radiation emitted from the radiation source,

Is the detection efficiency of the rotation modulated collimator 9 detector according to the energy.

Is the probability that the radiation emitted from the radiation source located at (x, y, z) at any one point in the three-dimensional Cartesian coordinate system at the solid angle enters the front side of the mask, and the radiation entering the mask located in front of the mask passes through the open space to the detector. The probability of reaching

The probability of blocking radiation in the mask

The sum of the two probabilities is 1. At this time, even if the photons arrive in the non-open space of the mask, the rate of radiation transmission according to the material and energy is

Is,

Means the number of background particles.

)을 기초로 바틀릿(Bartlett) 변환을 포함하는 분산 안정화 변환을 수행할 수 있다. [수학식 1]과 같이 프아송 분포로 얻어지는 변조 패턴의 관측값에서, 통계적 특성에 따라 프아송 잡음의 분산은 프아송 분포의 평균값(

)과 동일할 수 있다. 따라서, 분산 안정화 변환 단계(S1)에서는 이러한 평균과 분산의 연관성을 제거하기 위해 분산 안정화 변환을 수행할 수 있으며, 특히, 바틀릿(Bartlett) 변환을 사용하여 제1 검출 계수의 제1 변조 패턴(

)을 획득할 수 있다. 바틀릿(Bartlett) 변환은 다음의 [수학식 2]를 통해 수행될 수 있다.In the dispersion stabilization transformation step (S1), the imaging apparatus may calculate the first detection coefficient by performing dispersion stabilization transformation on the modulation pattern acquired by the rotation modulation collimator. The variance stabilization conversion step (S1) comprises a modulation pattern (

Can be performed on a distributed stabilization transformation including Bartlett transformation. In the observation of the modulation pattern obtained by the Poisson distribution as shown in [Equation 1], the variance of the Poisson noise according to the statistical characteristics is obtained from the average value of the Poisson distribution.

May be the same as). Therefore, in the dispersion stabilization transformation step S1, a dispersion stabilization transformation may be performed to remove the association between the mean and the variance, and in particular, the first modulation pattern of the first detection coefficient (Bartlett transformation) may be used.

) Can be obtained. Bartlett transformation can be performed through Equation 2 below.

[Equation 2]

)은 근사 평균(

)이

이고, 분산이 1인 가우시안 분포로 근사할 수 있다. The first modulation pattern (transformed via Bartlet transform)

) Is an approximate mean

)this

And a Gaussian distribution with a variance of 1.

)에 가중치

를 적용하여, 회전각도 i 에서의 카운트 값을 패치기반 가중평균으로 필터링 할 수 있으며, 이를 통해 제2 검출 계수의 집합인 제2 변조 패턴(

)을 획득할 수 있다. 제2 변조 패턴은 하기의 [수학식 3]을 통해 획득될 수 있다.In the noise removing step S3, after performing the dispersion stabilization step S1 in the imaging apparatus, the NLM algorithm may be performed on the first detection coefficient to calculate the second detection coefficient of the modulation pattern from which the noise is removed. Can be. The noise removing step S3 may include the first modulation pattern obtained in the dispersion stabilization conversion step S1.

Weight on)

By applying, it is possible to filter the count value at the rotation angle i by the patch-based weighted average, through which the second modulation pattern (a set of second detection coefficients)

) Can be obtained. The second modulation pattern may be obtained through Equation 3 below.

[Equation 3]

는 조건에 따라 상이하게 적용될 수 있으며 조건은 다음과 같다.At this time, the weight

May be applied differently depending on the conditions, and the conditions are as follows.

는 [수학식 4]를 따를 수 있다.According to the conditions

Can follow Equation 4.

[Equation 4]

는 변조 패턴(

)에서 기준이 되는 각도 i를 중심으로 하는 패치,

는 변조 패턴에서의 다른 각도를 의미하며

또한 i와 동일한 범위의 값을 가질 수 있다. 또한,

는 대역폭(bandwidth)이며,

는 각도 k를 중심으로 하는 패치이다. 일 예로, i가 10 도 일 때

는

,

는 80도 일 때

는

으로 나타내며, 이때 가중치(

)는

으로 표기되어 적용될 수 있다.here

Is the modulation pattern (

A patch centered on the angle i as a reference,

Means different angles in the modulation pattern

It may also have a value in the same range as i. Also,

Is the bandwidth,

Is a patch centered at angle k. For example, when i is 10 degrees

Is

,

Is at 80 degrees

Is

Where the weight (

)

It can be applied as indicated.

를 선택적으로 적용할 수 있다. 예를 들어,

를 적용하지 않은 가중치(

)를 [수학식 3]에 바로 적용할 수 있으며, 제2 변조 패턴(

)의 디테일을 유지하기 위해

를 초과하는 가중치(

)를 적용할 수 있다. 또한, 노이즈 제거 단계(S3)에서는 비지역 평균(NLM) 필터링을 적용함에 따라 추후 획득하게 되는 영상에서 아티팩트와 같은 노이즈가 효과적으로 제거될 수 있다. 한편, 가중치(

)를 계산하는 과정에서 제1 변조 패턴(

)에 평균 필터(mean filter)를 추가적으로 적용할 수 있다.When non-local mean (NLM) filtering is applied in the noise reduction step (S3),

Can be optionally applied. E.g,

Weight without

) Can be directly applied to Equation 3, and the second modulation pattern (

To keep the details of

Weights exceeding

) Can be applied. In addition, in the noise removing step S3, noise, such as an artifact, may be effectively removed from an image obtained later by applying NLM filtering. On the other hand, weights (

In the process of calculating the first modulation pattern (

) Can be additionally applied to the mean filter.

Each character described above may be equally applied to the following. In this case, the modulation pattern modeled as the Poisson distribution is approximated in a form similar to the Gaussian distribution as shown in Equation 5 through the dispersion stabilization transformation step S1 and the noise removing step S3.

[Equation 5]

)의 근사 평균(

)은

이다. As such, the modulation pattern having the transformed distribution cannot use the baseline algorithm based on the existing Poisson distribution, and a new imaging algorithm may be required. At this time, the second modulation pattern (

Approximation mean of

)silver

to be.

)에 최대 가능성 기대값 최대화(MLEM) 알고리즘을 적용할 수 있다. 상세하게, 영상 재구성 단계(S5)는 제2 검출 계수에 최대 가능성 기대값 최대화 알고리즘을 적용하여 방사선원의 위치 정보를 검출하고, 방사선원의 위치 정보를 기초로 재구성된 방사선원의 영상을 출력할 수 있다. 영상 재구성 단계(S5)에서는 전술한 바와 같이 분산 안정화 변환 단계(S1) 및 노이즈 제거 단계(S3)를 통해 대체적인 가우시안 분포를 나타내는 제2 변조 패턴(

)에 종래의 Baseline 알고리즘과는 상이한 영상화 알고리즘이 적용될 수 있다. The image reconstruction step S5 may include a second modulation pattern in which noise is removed from the imaging apparatus.

) Can be applied to the maximum likelihood expectation maximization (MLEM) algorithm. In detail, in the image reconstruction step S5, the position information of the radiation source may be detected by applying the maximum likelihood expected value maximization algorithm to the second detection coefficient, and the image of the reconstructed radiation source may be output based on the position information of the radiation source. In the image reconstruction step S5, as described above, a second modulation pattern (e.g., a Gaussian distribution) representing a general Gaussian distribution is obtained through the dispersion stabilization conversion step S1 and the noise removal step S3.

), An imaging algorithm different from the conventional Baseline algorithm may be applied.

)로부터 방사선원의 분포를 영상화하게 되며, 분산 안정화 변환 및 노이즈 필터링이 수행되지 않은 상태로 최대 가능성 기대값 최대화(MLEM) 알고리즘을 수행하여 방사선원의 분포(

)에 대하여 반복 알고리즘을 획득할 수 있다. In particular, the existing baseline algorithm has a problem in that noise such as a number of artifacts is generated in the imaging result when the radiation emitted from the radiation source is weak or the radiation is strong, but the detection coefficient counted by the detector due to a shield or an obstacle is small. . In addition, the existing baseline algorithm uses a modulation pattern obtained from a rotation modulation collimator (

Image of the radiation source, and perform a maximum likelihood expectation (MLEM) algorithm without dispersion stabilization transformation and noise filtering.

We can obtain an iterative algorithm for.

)에 최대 가능성 기대값 최대화(MLEM) 알고리즘을 적용하여 하기의 [수학식 6]과 같은 반복 알고리즘을 획득할 수 있다.In contrast, in the image reconstruction step S5 of the present invention, the second modulation pattern obtained in the noise removing step S3 (

) Can be obtained by applying a maximum likelihood expectation maximization (MLEM) algorithm to Equation 6 below.

[Equation 6]

는 평균적으로 얻어지는 광자 수이다. 이때,

는 픽셀 j에서 방출된 방사선이 회전각도 i에서 검출될 확률이며, 검출기의 검출효율 및 입체각과 검출확률을 고려해 사전에 계산할 수 있다. 또한,

은 근사 평균

의 도함수이다. [수학식 6]의 수식을 반복하는 반복 알고리즘을 통해 방사선원의 분포(

)를 추정할 수 있다. here,

Is the number of photons obtained on average. At this time,

Is the probability that the radiation emitted from the pixel j is detected at the rotation angle i, and can be calculated in advance in consideration of the detector's detection efficiency, solid angle, and detection probability. Also,

Is an approximate mean

Is a derivative of. Distribution of the radiation source through an iterative algorithm that repeats the equation in [Equation 6]

) Can be estimated.

로 나타낼 수 있다. 특히, 픽셀 j에서 방출된 방사선이 회전 변조 시준기(9)의 회전각도 i에서 검출될 확률을

로 나타낼 수 있다. 유사하게, 픽셀 j에서 방출된 방사선이 회전 변조 시준기(9)의 회전각도

에서 검출될 확률은

로 나타낼 수 있다. [수학식 6]의 분모와 분자에서 인덱스를 구분하기 위해 i와

가 구분되어 사용되었다. 반복 알고리즘을 통해 획득한 방사선원의 분포를 나타낸 예시를 도 8에 나타내었다. Specifically, the field of view (FOV) is divided into 21 * 21 (441) pixel units to determine the distribution of radiation of the radiation source at each pixel.

It can be represented as. In particular, the probability that the radiation emitted from pixel j will be detected at rotation angle i of rotation modulation collimator 9

It can be represented as. Similarly, the radiation emitted from pixel j causes the angle of rotation of the rotation modulation collimator 9

The probability to be detected at

It can be represented as. To distinguish the indices in the denominator and numerator of Equation 6, i and

Are used separately. An example of the distribution of the radiation source obtained through the iterative algorithm is shown in FIG. 8.

Hereinafter will be described an experimental example for comparing the results of the radiation source position detection method according to the present invention with each control.

Experimental Example

In this experimental example, the position of the radiation source was detected in (x, y, z), which is a three-dimensional rectangular coordinate system. The location of the radiation source is (25, 25, 500) cm and it is assumed that there is a Ba-133 dotted line that emits gamma rays of 0.356 MeV. The imaging algorithm was implemented using Matlab R2014b (Mathworks, USA).

) 사이즈는 3, 윈도우 사이즈는 360이고,

는 10^-0.8,

는 0.9이다. On the other hand, Figure 3 is a schematic diagram showing a mask of the rotation modulation collimator according to an embodiment of the present invention, referring to Figure 3, both the front and rear mask of the rotation modulation collimator is made of lead having a thickness of 1.27cm, six slits Has Based on the midpoint (o), the diameter (R) of the mask is 60 mm, the width (w) of each slit is 4 mm, the pitch is 8 mm, and each slit height (h _s ) fits a circle with a diameter (r) of 50 mm. It was. In addition, the field of view was calculated by dividing into 21 * 21 (441) pixels. The modulation pattern was obtained when the radiation of the radiation source was 0.2 mCi, and the patch (NLM) filter (

) Size is 3, window size is 360,

Is 10 ^-0.8 ,

Is 0.9.

<Experiment Result>

) 및 변조 패턴의 평균값(mean

), 도 5는 측정된 제1 변조 패턴(

) 및 제1 변조 패턴의 평균값(mean

), 도 6은 제1 변조 패턴에 가우시안 필터를 적용한 변조 패턴(

) 및 제1 변조 패턴의 평균값(mean

), 도 7은 제2 변조 패턴(

) 및 제1 변조 패턴의 평균값(mean

)을 나타낸다. 4 to 7 show modulation patterns according to photon measurement for each condition. In detail, FIG. 4 shows the measured modulation pattern (

) And the mean of the modulation pattern (mean

5 shows the measured first modulation pattern (

) And the mean value of the first modulation pattern

6 illustrates a modulation pattern in which a Gaussian filter is applied to a first modulation pattern (

) And the mean value of the first modulation pattern

), FIG. 7 shows a second modulation pattern (

) And the mean value of the first modulation pattern

).

)에서 상이한 방법으로 노이즈를 각각 제거한 측정 결과가 도 6 및 도 7에 나타난다. 도 6에서는 가우시안 필터를 적용하였고, 도 7에서는 비지역 평균(NLM) 필터를 적용하여 각각의 측정 결과를 비교하였다.In particular, the first modulation pattern of FIG.

In Fig. 6 and Fig. 7, the measurement results in which the noises are removed in different manners) are shown in Figs. In FIG. 6, Gaussian filters were applied, and in FIG. 7, non-local mean (NLM) filters were applied to compare the respective measurement results.

)의 단일 각도에서 검출된 광자의 평균값과 큰 차이를 나타내는 각각의 값들이 필터링되어 제1 변조 패턴(

)을 획득한 것을 확인할 수 있다. 도 4에서는 프아송 분포로 모델링 된 변조 패턴의 평균값(mean

)과 각도에 따라 검출된 변조 패턴의 각 광자의 측정값(

)의 차이가 평균값에 따라 달라지는 것을 확인할 수 있다. 즉, 측정되는 회전 각도에 따라 노이즈 크기가 상이하게 나타날 수 있다. 이러한 차이는 프아송 분포의 특성에 의한 것으로, 분산 안정화 변환 단계(S1)를 통해 도 4에 나타난 이러한 특성을 제거하여 도 5의 결과가 산출될 수 있다. 4 and 5, through the dispersion stabilization conversion step (S1) the initial modulation pattern (

Each value representing a large difference from the average value of the photons detected at a single angle of is filtered to obtain a first modulation pattern (

You can check that you have obtained. 4 shows the mean value of the modulation pattern modeled as the Poisson distribution.

) And the measured value of each photon of the detected modulation pattern according to the angle (

You can see that the difference of) depends on the mean value. That is, the noise level may be different depending on the rotation angle to be measured. This difference is due to the characteristics of the Poisson distribution, and the result of FIG. 5 may be calculated by removing such characteristics shown in FIG. 4 through the dispersion stabilization conversion step S1.

)과 평균값(mean

)의 차이가 도 4보다 감소한 것을 확인할 수 있다. 상세하게, 도 4의 변조 패턴(

)의 약 140도 지점에서는 약 14 이상으로 검출된 광자가 확인되지만, 분산 안정화 변환 단계(S1) 수행 이후 획득한 제1 변조 패턴(

)의 유사 지점에서는 약 8 에 가까운 광자가 검출된 것을 확인할 수 있다. 이와 같이 분산 안정화 변환 단계(S1)는 평균과 분산의 연관성을 제거할 수 있다. Referring to Figure 5, the overall measured value of the modulation pattern (

) And mean

It can be seen that the difference of) decreases from FIG. 4. Specifically, the modulation pattern of FIG.

At about 140 degrees, photons detected at about 14 or more are identified, but the first modulation pattern (A) obtained after performing the dispersion stabilization conversion step (S1)

In the similar point of), it can be seen that photons of about 8 were detected. As such, the dispersion stabilization transformation step S1 may remove the correlation between the mean and the variance.

)과 제1 변조 패턴에 가우시안 필터를 적용한 패턴의 평균값의 변화는 크게 변하지 않음을 알 수 있다. 다만, 제1 변조 패턴에 가우시안 필터를 적용함으로써 단일 각도에서 카운트 되는 광자 수가 검출된 광자수의 평균값에 밀집된 형태로 나타나는 것을 확인할 수 있다.Meanwhile, referring to FIGS. 5 and 6, the first modulation pattern (

) And the average value of the pattern in which the Gaussian filter is applied to the first modulation pattern do not change significantly. However, it can be confirmed that by applying a Gaussian filter to the first modulation pattern, the number of photons counted at a single angle appears in a dense form at an average value of the detected number of photons.

)과 제2 변조 패턴(

)의 평균값의 변화는 크게 변하지 않음을 알 수 있다. 반면, 제1 변조 패턴(

)과 제2 변조 패턴(

)의 동일한 회전 각도에 대하여 검출되는 광자의 평균값에 대한 밀집도가 상이해진 것을 확인할 수 있다.5 and 7, the first modulation pattern of the detected photons (

) And the second modulation pattern (

It can be seen that the change in the mean value of) does not change significantly. On the other hand, the first modulation pattern (

) And the second modulation pattern (

It can be seen that the density of the average value of the photons detected with respect to the same rotation angle of) is different.

)의 평균값을 기준으로 약 -4 내지 +4 범위의 변조 패턴이 확인되며, 도 7에서는 각각의 제2 변조 패턴(

) 의 평균값을 기준으로 약 -3 내지 +3 범위의 변조 패턴이 확인된다. 이처럼, 노이즈 제거 단계(S3)를 통해 제1 변조 패턴(

)에서 노이즈가 제거되어 제2 변조 패턴(

)을 획득할 수 있다. In detail, in FIG. 5, each first modulation pattern (

A modulation pattern in the range of about -4 to +4 is identified based on the average value of), and in FIG. 7, each second modulation pattern (

A modulation pattern in the range of about −3 to +3 is identified based on the average value of. In this way, the first modulation pattern (

) Is removed from the second modulation pattern (

) Can be obtained.

8 is a graph illustrating a method of detecting a radiation source position using a rotation modulation collimator and an imaging result of each control group according to an exemplary embodiment of the present invention.

)에 Baseline 알고리즘을 적용한 도 4의 영상화 결과, 도 8(b)는 도 5의 제1 변조 패턴(

)의 영상화 결과, 도 8(c)는 도 6의 제1 변조 패턴(

)에 가우시안 필터를 적용한 영상화 결과, 도 8(d)는 도 7의 제2 변조 패턴(

)의 영상화 결과를 각각 나타낸다. In detail, FIG. 8 (a) shows a modulation pattern (

As a result of the imaging of FIG. 4 applying the baseline algorithm to FIG. 4, FIG. 8B illustrates the first modulation pattern of FIG.

FIG. 8C shows the first modulation pattern of FIG. 6.

FIG. 8 (d) shows the second modulation pattern (FIG. 7) of the Gaussian filter.

Each of the imaging results.

In FIGS. 8A to 8D, positions with red circles represent positions where actual radiation sources exist, and positions of clear black boxes represent positions of radiation sources estimated through imaging. The reason for the symmetrical location of the radiation source is due to the structural limitation that the shape of the open space is the same with respect to 180 ° as the mask of the rotation modulation collimator rotates.

Meanwhile, it can be seen from FIG. 8 (a) to (d) that light black boxes are generated at positions other than the actual radiation source. This is an artifact that incorrectly assumes that there is a radiation source where there is no radiation source. In particular, the result of FIG. 8 (c) is light black, unlike the result of the modulation pattern obtained by performing Gaussian filtering on the first modulation pattern of FIG. 6. It can be seen that a number of boxes are generated to detect a large number of artifacts.

As such, it can be seen from the comparison of FIGS. 8A to 8D that the light black box is the least identified in FIG. 8D. In order to quantitatively evaluate the results of FIG. 8, a peak signal to noise ratio (PSNR) was performed for each imaging result. Equation of PSNR performed in the present invention is as follows.

[Equation 7]

은 픽셀의 수를 나타내고,

는 실재 방사선원의 분포를,

는 추정한 방사선원의 분포를 나타낸다. here

Represents the number of pixels,

Is the distribution of the actual radiation source,

Denotes the distribution of the estimated radiation source.

9 is a graph showing results of peak signal to noise ratio (PSNR) in radiation sources of various intensities.

In detail, FIG. 9 is a graph of the average obtained by repeating the same PSNR experiment 100 times in each radiation source assuming that the radiation emitted from the radiation source is in the range of 0.1 to 1.4 mCi.

Referring to FIG. 9, it can be seen that the result of the method for detecting the position of the radiation source using the rotation modulation collimator according to the present invention exhibits the highest value in all regions of the radiation intensity emitted from the applied radiation source. In particular, it can be seen that the PSNR is improved in comparison with the results of other algorithms in the region of low radiation.

In detail, the PSNR of the baseline algorithm of FIG. 4 is 41.30 (± 0.43) dB at the point where the radioactivity of the radiation source is 0.2 mCi, and increases to 44.17 (± 0.64) dB by the dispersion stabilization conversion of FIG. 5. In addition, when the noise is removed through the Gaussian filter after the dispersion stabilization transformation of FIG. 6, the measurement is 36.82 (± 0.25) dB, and the result of the radiation source position detection method using the rotation modulation collimator of FIG. It can be seen that (± 0.86).

As described above, the method for detecting the position of the radiation source according to the present invention, which solves the problem of artifacts when the intensity of the radiation source is weak, results in imaging quality in a region where the intensity of the radiation source is weaker than the result of performing only the existing baseline algorithm and some steps of the present invention. It can be seen that this provides further improved results.

On the other hand, according to the method of obtaining the position of the radiation source and the modulation pattern, it is attached [Table 1] containing the data indicating the results of the different PSNR at the radiation intensity of 0.2 mCi, through which the radiation source position detection method of the present invention various conditions It can be seen that also shows a high PSNR result.

Sailor's Position (x, y, z) cm (25,25,500) cm (0,40,500) cm (-40, -30,500) cm (-10,5,500) cm Measurements 41.30 (± 0.43) 46.31 (± 0.57) 44.53 (± 0.47) 42.72 (± 0.41) NLM + baseline 41.30 (± 0.43) 46.32 (± 0.57) 44.53 (± 0.47) 42.72 (± 0.42) variance stabilization (vs) 44.17 (± 0.64) 47.91 (± 0.62) 47.52 (± 0.66) 43.23 (± 0.45) vs + mean 43.28 (± 0.57) 46.69 (± 0.55) 45.72 (± 0.59) 43.31 (± 0.45) vs + gaussian 36.82 (± 0.25) 35.63 (± 0.17) 34.73 (± 0.14) 44.05 (± 0.48) vs + NLM 43.86 (± 0.61) 47.64 (± 0.60) 47.24 (± 0.65) 43.01 (± 0.44) proposed (vs + mean + NLM) 47.19 (± 0.86) 50.41 (± 0.85) 49.54 (± 0.76) 45.91 (± 0.66)

In summary, the radiation source position detection method using the imaging device 1 and the rotation modulation collimator of the present invention can quickly visualize the position of the radiation source using the rotation modulation collimator 9, and in particular, the obstacles caused by the small or obstruction of the radiation source. If is present, dispersion stabilization transformation may be performed on a modulation pattern modeled as a Poisson distribution, noise is removed through an NLM filter, and an MLEM algorithm may be applied to obtain a radiation source detection result of improved quality. In addition, it is possible to provide an improved image compared to the conventional imaging method in radiation of various intensities.

Although the present invention has been described in detail through the representative embodiments above, those skilled in the art to which the present invention pertains will appreciate that various modifications can be made without departing from the scope of the present invention. will be. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by all changes or modifications derived from the claims and the equivalent concepts as well as the claims below.

1: imaging device
11: converter
13: filtering section
15: detector
17: output unit
9: rotationally modulated collimator
S1: Variance Stabilization Conversion Step
S3: Noise Reduction Step
S5: Image Reconstruction Step

Claims (5)

A first detection coefficient and a first detection coefficient whose correlation between the mean and the variance of the modulation pattern are removed by performing dispersion stabilization transformation of the modulation pattern according to the number of photons detected by the rotation of a rotation modulation collimator (RMC) Calculating a first modulation pattern of coefficients;
The second detection coefficient from which the noise is removed is calculated by applying a weight calculated based on the similarity of the patch centering on the first angle and the second angle of the rotation modulation collimator to the first detection coefficient, and the first modulation pattern Calculating a second modulation pattern that is the set of the second detection coefficients by applying the weight to the filter and filtering the count value at the first angle with a patch-based weighted average; And
Reconstructing an image by estimating a distribution of the radiation source based on a probability that a calculated radiation source is detected according to at least one of the second detection coefficient and the second modulation pattern and the rotation angle of the rotation modulation collimator; A radiation source position detection method using a rotation modulated collimator.
The method of claim 1,
And calculating a weight by applying a predetermined filter when the number of detected photons is equal to or less than a threshold, the radiation source position detecting method using the rotation modulation collimator.
The method of claim 1,
Reconstructing the image,
And applying a maximum likelihood expectation maximization algorithm (MLEM) to derive an iterative algorithm based on the distribution of the second detection coefficients.
A transform unit for performing a dispersion stabilization transformation of the modulation pattern according to the number of photons detected by the rotation of the rotation modulation collimator to calculate a first detection coefficient and a first modulation pattern in which the correlation between the mean and the dispersion of the modulation patterns is removed;
The second detection coefficient from which the noise is removed is calculated by applying a weight calculated based on the similarity of the patch centering on the first angle and the second angle of the rotation modulation collimator to the first detection coefficient calculated by the converter. A filtering unit applying the weight to the first modulation pattern and filtering a count value at the first angle with a patch-based weighted average to calculate a second modulation pattern; And
The distribution of the radiation source is estimated by estimating the distribution of the radiation source based on a probability that the calculated radiation source is detected according to at least one of the second detection coefficient and the second modulation pattern calculated by the filtering unit and the rotation angle of the rotation modulation collimator. And a detector for acquiring position information.
The method of claim 4, wherein
And an output unit configured to reconstruct an image of the radiation source based on the position information acquired by the detector.

Images