KR20110121536A - Method and apparatus for super-resolution using wobble motion and psf in positron emission tomography - Google Patents

Abstract

PURPOSE: A super-resolution photographing device and a super-resolution photographing method using a PSF(Point Spread Function) and a wobble motion in a positron emission tomography are provided to implement an image with super-resolution based on the PSF according to the position of a PET(Positron Emission Tomography) image. CONSTITUTION: A signal classifying unit(110) classifies an input signal applied through the movement of a bed according to the position of a PET. A first image generator(120) generates a first image set by re-composing the classified input signal. A first image set is a low resolution image set. A parameter measuring unit(130) measures the PSF according to the first image set. A second image generator(140) generates second image information by applying super-resolution algorithm based on the PSF.

Description

TECHNICAL AND APPARATUS FOR SUPER-RESOLUTION USING WOBBLE MOTION AND PSF IN POSITRON EMISSION TOMOGRAPHY}

Embodiments of the present invention are a super-resolution imaging apparatus (hereinafter, referred to as an image generating apparatus) and a method using a wobble operation and the PSF in a positron tomography image generating a PET image using a super-resolution technique. It is about.

Positron emission tomography (PET) is widely used as one of the methods of nuclear medicine that can display physiological, chemical, and functional images of the human body in three dimensions by using radiopharmaceuticals that emit positrons.

The positron emission tomography (PET) is mainly used to diagnose various types of cancers and provides useful results in differential diagnosis, staging, recurrence evaluation, and treatment effect determination.

In addition, positron emission tomography (PET) can be used to obtain receptor images or metabolic images for heart disease, brain disease, and brain function evaluation.

Positrons emitted from radioactive isotopes consume all of their kinetic energy for a very short time after their release and combine with neighboring electrons to extinguish, which emits two extinction radiation (gamma rays) at an angle of 180 °.

A positron tomography (PET) scanner made of cylinders can detect two extinguished radiation emitted simultaneously. When the image is reconstructed using the detected radiation, a 3D tomography image of how much radiopharmaceuticals are collected in which part of the body can be represented.

However, in the process of acquiring the PET information, the image may be blurred due to positron range, non-colinearity, detector size, etc., and thus it may be difficult to obtain accurate results by generating a blurred or shaken image. .

An image generating apparatus according to an embodiment of the present invention aims to generate a clearer image.

In addition, an image generating apparatus according to an embodiment of the present invention aims to generate a super resolution image based on a point spread function (PSF) according to a position of a Positron Emission Tomography (PET) image.

According to an embodiment of the present invention, an image generating apparatus includes a signal classifier configured to classify an input signal based on a location of the PET through movement of a whole Positron Emission Tomography (PET) detector or a bed, and reconstruct the classified input signals. A first image generator for generating a first image set, a parameter measurer for measuring a point spread function (PSF) according to the first image set, and a super-resolution algorithm based on the PSF And a second image generator configured to generate second image information.

In addition, the image generating method according to an embodiment of the present invention is to classify the applied input signal according to the position of the PET through the movement of the entire PET (Positron Emission Tomography) detector or bed, reconstruct the classified input signal Generating a first image set, measuring a point spread function according to the first image set, and generating second image information by applying a super-resolution algorithm based on the PSF. It includes a step.

According to an embodiment of the present invention, a clearer image may be generated.

In addition, according to an embodiment of the present invention, a super resolution image may be generated based on a point spread function (PSF) according to a position of a Positron Emission Tomography (PET) image.

1 is a block diagram showing the configuration of an image generating apparatus according to a first embodiment of the present invention;
2 is a flowchart illustrating an image generating method according to a first exemplary embodiment of the present invention.
3 is a diagram illustrating a PET photographing scene in the image generating method according to the first embodiment;
4 is a diagram illustrating an input signal measuring process of a PET device generating an input signal with an image generating device according to a second embodiment of the present invention.
5 is a diagram illustrating a process of measuring a PSF using an image generating device according to a third embodiment of the present invention.
6 is a diagram for explaining a general form of parallax error.
7 illustrates a correlation between detector size and tangential blur.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and accompanying drawings, but the present invention is not limited to or limited by the embodiments.

On the other hand, in describing the present invention, when it is determined that the detailed description of the related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. The terminology used herein is a term used for appropriately expressing an embodiment of the present invention, which may vary depending on the user, the intent of the operator, or the practice of the field to which the present invention belongs. Therefore, the definitions of the terms should be made based on the contents throughout the specification.

1 is a block diagram illustrating a configuration of an image generating apparatus according to a first embodiment of the present invention.

Referring to FIG. 1, the image generating apparatus includes a signal classifier 110, a first image generator 120, a parameter measurer 130, and a second image generator 140.

The image generating apparatus applies an input signal through the movement of the entire positron emission tomography (PET) detector or a bed, generates a first image set based on the input signal classified according to the position of the PET detector, and generates the first image set. After the measurement of the point spread function (PSF) according to the image set, the improved second image information may be generated through the super resolution image technique.

A method of generating an improved image through a super resolution image technique using an image generating apparatus according to the first embodiment will be described below with reference to FIGS. 2 and 3.

2 is a flowchart illustrating an image generating method according to a first embodiment of the present invention, and FIG. 3 is a diagram illustrating a PET photographing scene in the image generating method according to the first embodiment.

The image generating apparatus is an apparatus for generating image information based on a super resolution algorithm using an input signal applied as the measurement target passes through the PET 310 as shown in FIG. 3.

In particular, the image generating apparatus is characterized in that the image is generated by applying the input signal measured due to the rotational motion, such as the wobble (wobble) movement of the entire PET detector or the patient lying bed.

The image generating apparatus applies an input signal applied through the movement of the entire PET (Positron Emission Tomography) detector or a bed using the signal classification unit 110 to generate image information by applying a super resolution algorithm to the position of the PET. (210).

Next, the image generating apparatus generates a first image set by reconstructing the classified input signals using the first image generator 120 (220).

In this case, the first image set is a set of low-resolution (LR) images. Accordingly, the apparatus for generating an image may refer to an image set including one or more image information, and later, may generate second image information by using first image information including one or more image information.

For example, the image generating apparatus may generate second image information, which is one piece of 128x128 image information, by using a first image set including four pieces of 64x64 image information.

4 is a diagram illustrating an input signal measuring process of a PET device generating an input signal with an image generating device according to a second embodiment of the present invention.

For example, as shown in FIG. 4, the image generating apparatus measures a plurality of reconstructed image sets according to the position of the PET device by measuring the input signal while the entire PET 310 detector is circularly moved due to the movement of a wobble or the like. Can be obtained.

In this case, since the image set applied to the image generating apparatus may be regarded as non-integer pixel shift, a super-resolution algorithm may be applied.

That is, the image generating apparatus moves the entire Positron Emission Tomography (PET) detector by a wobble operation, classifies the measured input signals according to the position of the PET device, and reconstructs the input signals to generate a plurality of first image sets. Can be generated.

In this case, since the first image information is translated in parallel to each other, this may have the same concept as taking a picture with a slight movement of the same object in the camera. Accordingly, the sharper image information may be generated by applying a super resolution algorithm using the first set of parallel images.

In other words, the image generating apparatus obtains an input signal through a wobble operation, and classifies and reconstructs the input signal according to the position of the PET device to generate a low-resolution image set as a first image set. .

In this case, the reconstruction algorithm that can be used in the image generating apparatus includes an analytical reconstruction algorithm such as filtered-backprojection (FBP) and an iterative reconstruction algorithm such as ordered-subset expectation maximization (OSEM). Various reconstruction algorithms can be used.

Next, the image generating apparatus according to the second embodiment measures a point spread function (PSF) according to the first image set by using the parameter measuring unit 130 (230).

5 is a diagram illustrating a process of measuring a PSF using an image generating device according to a third embodiment of the present invention.

In general, the image information measured by the PET device may cause blur due to positron range, non-colinearity, and detector size. For example, the PSF of a sample measured in a ring-shape or cylindrical PET system may vary by location.

As shown in FIG. 5, the image generating apparatus uses a PSF that is different for each position as a blur model of a super resolution technique by using a characteristic of changing the shape of the PSF according to the position of the PET detector. Can be obtained.

That is, since the image generating apparatus also changes the PSF of the reconstructed first image set according to the position, the image generating apparatus measures the PSF according to the position of the first image set and applies the super resolution algorithm to generate the image information. In this case, if the PSF is used as a kernel of a super-resolution algorithm according to the position of the PET, an improved image can be obtained.

Finally, the second image generator 140 of the image generating apparatus generates second image information by applying a super-resolution algorithm based on the PSF (240). In this case, the second image information according to the first embodiment of the present invention is information about a high-resolution (HR) image.

That is, the second image generator 140 of the image generating apparatus may finally generate more high-resolution second image information by using the PSF as a blur model.

In this case, the image generating apparatus may generate second image information which is high resolution image information by applying the following super resolution algorithm.

For example, the second image generator 140 may generate second image information by applying Equation 1 below with the super-resolution algorithm.

는 제1 영상 세트(1≤k≤p)이고, 상기 p는 상기 제1 영상 세트의 개수이고, 상기

는 n번째 제2 영상 정보이다.At this time,

Is a first image set (1 ≦ k ≦ p), and p is the number of the first image sets,

Is n-th second image information.

는 다운 샘플링(down-sampling), 블러링(blurring), 평행 이동(translation)을 포함하는 매트릭스 값이다.In addition,

Is a matrix value that includes down-sampling, blurring, and translation.

는 결과 영상을 부드럽게 만들어 주는 부분이다. In addition, C is a high-pass filter value,

Is the part that softens the resulting image.

Α denotes a smoothness parameter, β denotes a convergence parameter, and T denotes a matrix transpose.

) 수행 과정은 영상 정보를 선명(sharp)하게 만들어 주는 하이패스 필터 통과 과정이라 할 수 있다.For example, the previous part of α in Equation 1

) Is a high pass filter pass that makes the image information sharp.

) 수행 과정은 영상 정보를 매끄럽게(smooth) 만들어주는 과정으로 로우패스 필터 통과 과정이라 할 수 있다.On the other hand, according to a further embodiment of the invention, the part after the α (

) Is a process of smoothing image information and may be referred to as a low pass filter passing process.

) 수행 과정과 (

) 수행 과정 중 어느 과정에 더 가중치를 둘지 여부를 결정할 수 있다.Furthermore, the image generating apparatus adjusts α in Equation 1, thereby (

) Process and (

) Can determine whether to weight more of the process.

를 계속적으로 업데이트 하며, β를 조절하여 상기 업데이트 값의 크기를 결정할 수 있다. 예를 들어, 수학식 1의 β의 값이 기준치 이상 큰 경우, 상기 영상 생성 장치는 적은 반복 수행(iteration)으로도 원하는 결과를 얻을 수 있다.In addition, the image generating apparatus of Equation 1

Is continuously updated and the size of the update value can be determined by adjusting β. For example, when the value of β in Equation 1 is greater than or equal to the reference value, the image generating apparatus may obtain a desired result with little iteration.

In addition, the image generating apparatus uses a maximum likelihood expectation maximization (MLEM) algorithm as an embodiment when a negative value is not included in a value of the first image set, which is the low resolution image set. Second image information may be generated through iterative computation.

That is, when the negative value is not included in the value of the first image set, the second image generator 140 may generate second image information by applying a maximum likelihood expectation maximization (MLEM) algorithm of Equation 2. .

의 초기 값(

)과 저해상도 영상(

)가 모두 비음수(non-negative)인 성질을 갖고 있어야 할 필요가 있다. 한편, 상기

는 제1 영상 세트(1≤k≤p)이고, 상기

는 상기 제1 영상 세트의 개수이고, 상기

는 n번째 제2 영상 정보이다.Here, to generate the second image information

Initial value of

) And low resolution images (

) Must all have non-negative properties. Meanwhile, above

Is the first image set (1≤k≤p),

Is the number of the first image set, and

Is n-th second image information.

는 다운 샘플링(down-sampling), 블러링(blurring), 평행 이동(translation)을 포함하는 매트릭스 값이다.In addition,

Is a matrix value that includes down-sampling, blurring, and translation.

In addition, T means a matrix transpose.

As a result, the image generating apparatus may obtain a result by applying a weighted sum blurring model to blur the first image set, which is a high resolution sample, that is, the weight The value may correspond to the PSF of the PET.

At this time, the PSF varies depending on the position of the PET 310 (spatial variance), and as shown in FIG. 5, the PSF can be obtained by measuring a point source.

In this case, the image generating apparatus may measure the PSF changed by repeatedly measuring the point source.

In conclusion, since the relationship between the high resolution image and the low resolution image is important in the image generating apparatus, if the PSF of the low resolution image is known, the image generating apparatus may determine the PSF of the low resolution image when generating the high resolution image. By using, you can make a clearer picture.

Furthermore, one of the reasons why the resolution in PET can vary from spatial location to location is parallax error. 6 is a diagram for explaining a general form of the parallax error.

Referring to FIG. 6, it can be seen that the geometry of the parallax error is shown and the resolution may vary depending on the distance from the center of the system to the current position.

That is, the system center corresponds to A, and the event occurs in the system B. In this case, the probability that a signal is detected through an adjacent crystal becomes high, and data may be recorded in an Inaccurate LOR corresponding to C.

As such, the parallax error may cause the object to be blurred by moving the position of the object toward the center of the system, and this phenomenon may become worse as the distance from the center of the system is increased. In addition, another reason why the resolution in PET may vary is tangential blur. 7 is a diagram for explaining a correlation between detector size and tangential blur.

Referring to FIG. 7, D1 and D2 represent detectors, each of which corresponds to D1 and D2 shown in FIG. 6. In particular, in annihilation position P, if two gamma rays are emitted, the probability that each of the emitted gamma rays are simultaneously measured at detectors D1 and D2 may be proportional to the solid angle.

Thus, through FIG. 7, the solid angle for each of the detectors D1 and D2 can be described at any position between the detectors D1 and D2.

The intermediate position between the detectors D1 and D2 has a narrow distribution of solid angles, while the closer the detectors D1 and D2 are, the wider the distribution of solid angles is. Therefore, at the system center, the position can be estimated to be close to the actual position using only projection information.

However, the more uncertain the position is from the center of the system, the more difficult it is to estimate the position closer to the actual position using only projection information. In other words, as the position moves away from the center of the system, the resolution becomes worse, resulting in spatial resolution imbalances.

In order to solve this problem, the image generating apparatus calculates a blur kernel for each position by positioning a point source for each position of a high resolution image and acquiring data for each position according to the fourth embodiment of the present invention. can do.

In particular, the image generating apparatus may use Monte Carlo simulation, and in this case, the point source may be more precisely positioned than before.

In addition, when the image generating apparatus calculates a blur kernel for each position, the image generating apparatus may convert the image to an image through a reconstruction process. In particular, the image generating apparatus may know how the source is blurred when the source is positioned in each pixel of the high resolution image, and thus, the image generating apparatus may detect various blur phenomena such as parallax error and tangential blur problem due to detector size. Characteristics can be taken into account.

For example, each of the low resolution images is generated by down-sampling the blurred high resolution image after the high resolution image is blurred. However, Monte Carlo simulations can measure blur kernels at low resolution intervals rather than at high resolution intervals.

Accordingly, the image generating apparatus may provide a function of performing a blur process and a down sampling process in the super resolution (SR) method.

In other words, the image generating apparatus may convert the high resolution image into a low resolution image by blurring and down-sampling. In addition, the image generating apparatus may compensate for the tangential blur problem due to the position correction of the object and the detector size by using the blur kernel for the low resolution calculated in real time using Monte Carlo simulation.

In addition, in the present invention, the super resolution algorithm may have an ill-posed problem. In other words, when maximizing or minimizing a certain cost function (e.g., likelihood function for Poisson), there may be several solutions with the same cost function.

In order to solve this problem, the image generating apparatus may provide regularization and derive a unique solution through the normalization according to the fifth embodiment.

In particular, the image generating apparatus may use a total-variation normalization that has an edge preserving property that preserves a large gradient portion.

In addition, the image generating apparatus may calculate a solution limited to a positive number by adjusting several parameters.

For example, according to the normalization characteristic used in the MAP-EM, since a negative solution may be calculated, the image generating apparatus uses total-variance normalization to limit the solution to a value within a predetermined range or not to be negative. You can do it.

Embodiments according to the present invention can be implemented in the form of program instructions that can be executed by various computer means can be recorded on a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of the computer-readable recording medium include magnetic media such as a hard disk, a floppy disk, and a magnetic tape; optical media such as CD-ROM and DVD; magnetic recording media such as a floppy disk; Magneto-optical 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 described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

110: signal classifier 120: first image generator
130: parameter measuring unit 140: second image generating unit

Claims (19)

A signal classification unit classifying the input signal applied through the movement of the entire positron emission tomography (PET) detector or the bed according to the position of the PET;
A first image generator configured to reconstruct the classified input signals to generate a first image set;
A parameter measuring unit measuring a point spread function (PSF) according to the first image set; And
A second image generator for generating second image information by applying a super-resolution algorithm based on the PSF;
Image generating apparatus comprising a. The method of claim 1,
The first image set is a set of low-resolution images,
And the second image information is information about a high-resolution image. The method of claim 1,
The first image generator,
And generating the first set of images through at least one of an analytic reconstruction algorithm and an iterative reconstruction algorithm. The method of claim 3,
The analytical reconstruction algorithm is a filtered-backprojection (FBP) algorithm,
And the iterative reconstruction algorithm is an ordered-subset expectation maximization (OSEM) algorithm. The method of claim 1,
The second image generator,
And generating the second image information by using the PSF as a blur model. The method of claim 1,
The second image generator,
And generating second image information by applying the following equation with the super-resolution algorithm.

(remind

Is the first image set (1≤k≤p),

Is the number of the first image set, and

Is n-th second image information, and

Is a matrix value including down-sampling, blurring, and translation, and

Is the high-pass filter value, α is the smoothness parameter, β is the convergence parameter, and T is the matrix transpose.) The method of claim 1,
The second image generator,
And generating a second image information by applying a maximum likelihood expectation maximization (MLEM) algorithm.

(remind

Is the first image set (1≤k≤p),

Is the number of the first image set, and

Is n-th second image information, and

Is a matrix value including down-sampling, blurring, and translation, T is a matrix transpose, and

Is a regularization term that uses total-variation,

Is a regularization parameter that indicates the degree of regualarization.) A signal classification unit classifying the input signal applied through the movement according to the position of the PET while positioning a point source for each pixel position of the high resolution image through movement of the entire PET detector or a bed;
A first image generator configured to reconstruct the classified input signals to generate a first image set;
A parameter measuring unit measuring a point spread function (PSF) according to the first image set; And
A second image generator for generating second image information by applying a super-resolution algorithm based on the PSF;
Including,
The second image generator
Applying the super resolution algorithm, obtaining data corresponding to the point source for each pixel position, and calculating a blur kernel for each pixel position based on the obtained data
Video generation device. The method of claim 8,
The second image generator
The position of the object due to parallax error in the second image information based on the low resolution image converted by blurring and down-sampling the high resolution image and the blur kernel of the low resolution image calculated in real time by Monte Carlo simulation To compensate for one or more of the tangential blur problem by calibration and detector size
Video generator Classifying the input signal applied through the movement of the entire positron emission tomography (PET) detector or a bed according to the position of the PET;
Reconstructing the classified input signals to generate a first set of images;
Measuring a point spread function (PSF) according to the first image set;
Generating second image information by applying a super-resolution algorithm based on the PSF;
Image generation method comprising a. The method of claim 10,
The first image set is a set of low-resolution images,
And the second image information is information about a high-resolution image. The method of claim 11,
The eleventh video set is
An image generating method generated by at least one of an analytic reconstruction algorithm or an iterative reconstruction algorithm. The method of claim 10,
The analytical reconstruction algorithm is a filtered-backprojection (FBP) algorithm,
The iterative reconstruction algorithm is an ordered-subset expectation maximization (OSEM) algorithm. The method of claim 11,
The second video information,
An image generating method generated by using the PSF as a blur model. The method of claim 10,
The generating of the second image information may include:
And generating second image information by applying the following equation with the super-resolution algorithm.

(remind

Is the first image set (1≤k≤p),

Is the number of the first image set, and

Is n-th second image information, and

Is a matrix value including down-sampling, blurring, and translation, and

Is the high-pass filter value, α is the smoothness parameter, β is the convergence parameter, and T is the matrix transpose.) The method of claim 8,
The generating of the second image information may include:
And when the negative value is not included in the value of the first image set, generating second image information by applying a maximum likelihood expectation maximization (MLEM) algorithm.

(remind

Is the first image set (1≤k≤p),

Is the number of the first image set, and

Is n-th second image information, and

Is a matrix value including down-sampling, blurring, and translation, T is a matrix transpose, and

Is a regularization term that uses total-variation,

Is a regularization parameter that indicates the degree of regualarization.) Positioning a point source for each pixel position of a high resolution image through movement of a bed or a positron emission tomography (PET) detector;
Classifying the input signal applied through the movement according to the position of the PET;
Reconstructing the classified input signals to generate a first set of images;
Measuring a point spread function (PSF) according to the first image set;
Generating second image information by applying a super-resolution algorithm based on the PSF; And
Applying the super resolution algorithm to obtain data corresponding to the point source for each pixel position, and calculating a blur kernel for each pixel position based on the obtained data;
Image generation method comprising a. The method of claim 17,
Generating the second image information
Based on the blur kernel of the low resolution image calculated in real time by Monte Carlo simulation and the low resolution image converted by blurring and down-sampling the high resolution image, the position of the object due to parallax error in the second image information Image generation method that compensates for one or more of the tangential blur problem due to correction and detector size. A computer readable recording medium having recorded thereon a program for performing the method of claim 10.

Images