CN103393434B - Method for obtaining system response model of positron emission tomography and method for image reconstruction - Google Patents

Abstract

The invention discloses a method and device for acquiring a system response model and image reconstruction of positron emission tomography. An image reconstruction method for positron emission tomography, comprising: acquiring depth effect response information of gamma photons incident on a crystal strip array; according to the depth effect response information, generating response lines determined by back-to-back gamma photon incident crystal strip pairs a depth effect response model; generate a system response model according to the depth effect response model; and perform image reconstruction according to the system response model.

Description

Method for Obtaining Response Model and Image Reconstruction of Positron Emission Tomography System

technical field

This patent relates to the field of nuclear medicine detection and imaging, and in particular to a method for accurately obtaining response models of PET systems with various geometric shapes and image reconstruction.

Background technique

Positron emission tomography (PET, hereinafter referred to as PET) is an important tomographic imaging device in the field of nuclear medicine, and has been widely used in diagnosis and research in the medical field. PET can effectively detect the spatio-temporal distribution of drugs labeled with radiotracer atoms injected into living organisms. Because the metabolism of certain abnormal tissues (such as tumors) is stronger than that of normal cells, the distribution of tracers in these tissues is also more than that of normal cells. Therefore, the distribution of these tracers detected by PET provides a basis for carrying organisms. Images of functional information (such as information about tumors) provide a powerful reference for the diagnosis and research of early tumors.

PET can be in various geometric shapes (such as flat plate, ring, etc.), which are formed by different arrangements of each detector module in PET, but the detection principle is the same. Before scanning, the organism is injected with a tracer containing positron radionuclides, and each positron in the tracer can annihilate with a negative electron in the organism to generate two back-to-back gamma photons, that is, a gamma photon pair , the two gamma photons pass through the biological tissue, hit a pair of detectors in the positron electron scanner, and perform a series of electronic responses, and then transmit the signal to the computer for recording. In this way, all the gamma photon pairs are recorded, and then after image reconstruction, an image reflecting the distribution of the radioactive tracer can be obtained. The main components in a PET detector consist of a scintillation crystal, a photomultiplier tube and front-end electronics. When gamma photons enter the scintillation crystal, the scintillation crystal converts the high-energy photon signal into low-energy photons, and the low-energy photons are converted into electrons through the photomultiplier tube and amplified, and output by the front-end electronics. In this way, the electrical signals simultaneously output from the ends of the two crystal strips record the information about the above-mentioned gamma photon pair, that is, an annihilation event. Usually, some simple models are used to describe the above-mentioned response situation, and the response of the PET system can be easily analyzed. Commonly used models include wire model, surface model and volume model, among which the wire model is the most commonly used. Since the two γ photons of the annihilation event are back-to-back, it can be approximated as a straight line. The back-to-back γ photons generated by the annihilation event are incident on the two crystal strips, and the line connecting the two crystal strips is called the line of response. , LOR, hereinafter referred to as the LOR line), as shown in Figure 1. In Fig. 1, LOR1 and LOR2 are two different LOR lines, LOR1 is the LOR line obtained by detector crystal strips a and b conforming to detection, and LOR2 is the LOR line obtained by detector crystal strips c and d conforming to detection.

Signals recorded directly with detectors can be organized in different data organizations. Such as List mode (list mode) and Sinogram (sinogram). The list mode data composition signal contains the number of the crystal bar where the two annihilation γ photons are incident, energy information and time information. The Sinogram only contains the position information and energy information of two annihilation gamma photon detections, but its recording method is to use the LOR line to determine the direction of its back-to-back annihilation photon, and its horizontal axis is the angle θ between the LOR line and the horizontal axis in the image space , the vertical axis is the distance r from the origin of the image space to the LOR line, and the position of each LOR line can be determined by (r, θ). Image reconstruction, one of the key technologies in PET imaging, is to use the above-mentioned coincident signals (such as Sinogram or List mode) and use the corresponding reconstruction algorithm to reconstruct the distribution image of the tracer in the organism. During the reconstruction process, the PET system response must be modeled in terms of the PET system geometry and detection characteristics. The accuracy of the system response model has a great relationship with the quality of the reconstructed image.

When building a system response model, the projection space and the image space are involved. The projection space is the space where the signal we detect is located, and it can be regarded as a projection of the image space to the detector. When establishing a model for the system response, the corresponding coordinate systems are established for the projection space and the image space at the same time, and the system response model based on these coordinate systems is established.

During the detection process, the main factors that affect the quality of the image are: the penetration and scattering between the crystal strips when the gamma photon interacts with the crystal strips. When the gamma photon is incident on a certain crystal strip of the detector at a certain angle, when it interacts with the material of the crystal strip, it will penetrate the incident crystal strip to the adjacent crystal strip, so that the penetrated crystal strip (including Both the incident crystal strip and the penetrating adjacent crystal strip) respond, this effect is called penetration, also known as depth effect (depth of interaction, DOI, hereinafter referred to as DOI effect). As shown in Figure 2A, assuming that an annihilation event occurs at a certain point on LOR1, it should be a real response line, when the γ photons incident along the direction of LOR1 are incident on a pair of crystal strips a1 and a2 at a certain angle, they penetrate into b1 respectively , b2 and attenuated absorption, so that the b1, b2 crystal strip pair also generates signals, so when reconstructing the image, it will be considered that annihilation events have occurred on both the LOR1 line and the LOR2 line, resulting in blurred image effects and reduced resolution, as shown in the figure 2B. Similarly, the scattering of gamma photons between crystals can also cause the resolution of the image to decrease. As shown in Figure 3A, gamma photons are scattered at the upper detector a1 crystal strip to b1 crystal strip, and penetrate at the lower end Acting on the b2 crystal strip, the b1, b2 crystal strip pair also responds, resulting in blurred images and reduced resolution, as shown in Figure 3B.

In order to solve this image blurring effect, the DOI problem can be improved on the hardware, and the crystal strip is designed to be double-layered or multi-layered, but this method is very expensive. If the DOI effect can be accurately simulated in the PET system response model and used in the image reconstruction process to restore the distribution of the image more realistically, it is an effective method. This method is low in cost, easy to modify, and has a high utilization rate of hardware. How to establish a system response model more accurately and reconstruct high-precision images has become a hot issue in the field of PET research today. At present, the methods of obtaining the system response model are generally divided into three types: analytical calculation method, Monte Carlo simulation method, and experimental measurement method.

However, the current system response model acquisition method mainly has the following defects: first, the calculation intensity is high, time-consuming, and the storage capacity is large, time-consuming and labor-intensive; second, the accuracy is poor, so that high-quality images cannot be obtained; third , poor versatility, many methods can only be applied to a detector geometry, and when the shape of the detector is changed without changing the crystal strip specification, it needs to be recalculated.

Contents of the invention

In view of the above, the object of the present invention is to provide a method and device for obtaining a PET system response model and image reconstruction with high image quality, strong versatility, and can save time and storage costs.

One aspect of the present invention discloses a method for obtaining a system response model of positron emission tomography, the method comprising:

Obtain the depth effect response information of the gamma photon incident on the crystal strip array;

According to the depth effect response information, the depth effect response model that generates the response line determined by the back-to-back gamma photon incident crystal strip pair time; and,

A system response model is generated based on the depth effect response model.

Another aspect of the present invention discloses an image reconstruction method of positron emission tomography, the method comprising:

Obtain the depth effect response information of the gamma photon incident on the crystal strip array;

According to the depth effect response information, it conforms to the depth effect response model for generating the response lines determined by the back-to-back gamma photon incident crystal strips;

generating a system response model based on the depth effect response model; and,

Image reconstruction is performed according to the system response model described.

Another aspect of the present invention discloses a device for acquiring a system response model and image reconstruction of positron emission tomography, the device comprising:

An acquisition module, configured to acquire depth effect response information of gamma photons incident on the crystal strip array;

The coincidence module is used to conform to generate a depth effect response model of a response line determined by back-to-back gamma photon incident crystal strips according to the acquired depth effect response information;

a calculation module, configured to calculate and generate a system response model according to the depth effect response model; and,

The reconstruction module is used for performing image reconstruction according to the system response model.

The method and device for obtaining the PET system response model and image reconstruction provided by the present invention, by obtaining the DOI response information of the gamma photon incident on the crystal strip array, conforms to the DOI response model of the LOR determined by the back-to-back gamma photon incident crystal strip pair, and will It is applied to the reconstruction process to solve the image blur caused by the DOI effect, obtain high-quality images, and has good portability, and can save time and storage costs, as long as the crystal strips used in PET have the same specifications, No matter how the geometry of PET changes, there is no need to repeatedly simulate the DOI response information generated by the γ-photon incident crystal strip, and the process method is simple, and the system response model can be generated in real time during the reconstruction process, which saves time and cost and storage costs.

Description of drawings

The above and other features and advantages of the present disclosure will become more apparent by describing in detail example embodiments thereof with reference to the accompanying drawings.

Fig. 1 is the schematic diagram of LOR;

Figure 2A is a schematic diagram of the penetration of gamma photons between crystals;

Fig. 2B is an imaging effect diagram when gamma photons penetrate through crystals;

Figure 3A is a schematic diagram of the scattering of gamma photons between crystals;

Fig. 3B is an imaging effect diagram when gamma photons are scattered between crystals;

Fig. 4 is an overall flowchart of a method for obtaining PET system response and image reconstruction disclosed by the present invention;

Figure 5A is a schematic diagram of a gamma photon vertically incident on a crystal strip;

Figure 5B is a schematic diagram of the DOI effect when gamma photons penetrate two crystals;

Figure 5C is a schematic diagram of the DOI effect when gamma photons penetrate three crystals;

Fig. 6 is a schematic diagram of the DOI effect when simulating multiple gamma photons incident at the same incident angle;

Figure 7A is a schematic diagram when the incident angle of gamma photons is an acute angle;

Fig. 7B is a schematic diagram when the incident angle of gamma photons is an obtuse angle;

Fig. 8 is a schematic diagram of the timing of gamma photons to the incident crystal strip;

Fig. 9 is a schematic diagram of fuzzy response produced by DOI effect;

Fig. 10 is the schematic diagram of the edge truncation effect of double flat panel detector;

Fig. 11A is the real structural diagram of the ring detector;

Figure 11B is a schematic diagram of the edge truncation effect of the ring detector;

FIG. 12 is a flowchart of an iterative reconstruction algorithm in an embodiment of the present invention;

Fig. 13 is a schematic diagram of an apparatus for acquiring a PET system response model and image reconstruction disclosed by an embodiment of the present invention.

Detailed ways

Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. In the drawings, the thickness of regions and layers are exaggerated for clarity. The same reference numerals in the drawings denote the same or similar structures, and thus their detailed descriptions will be omitted.

The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided in order to give a thorough understanding of embodiments of the invention. However, one skilled in the art will appreciate that the technical solutions of the present invention may be practiced without one or more of the specific details, or that other methods, components, materials, etc. may be employed. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the invention.

The present invention discloses a method for obtaining PET system response and image reconstruction, the overall process of which is shown in Figure 4, including the following steps:

Step S401: Obtain DOI response information of gamma photons incident on the crystal strip array;

Step S402: According to the DOI response information, it conforms to the DOI response model of the LOR determined by the pair of back-to-back γ-photon incident crystal strips;

Step S403: Generate a system response model according to the generated DOI response model;

Step S404: Perform image reconstruction according to the system response model.

The method and device for obtaining the PET system response model and image reconstruction provided by the present invention, by obtaining the DOI response information of the gamma photon incident on the crystal strip array, conforms to the DOI response model of the LOR determined by the back-to-back gamma photon incident crystal strip pair, and will It is applied to the reconstruction process to solve the image blur problem caused by the DOI effect. While obtaining high-quality images, it has good portability and can save time and storage costs, as long as the specifications of the crystal strips used in PET are the same. , no matter how the geometry of the PET changes, it is not necessary to repeatedly simulate the DOI response information generated by the γ-photon incident crystal strip, and the process method is simple, and the system response model can be generated in real time during the reconstruction process, which saves time costs and storage costs.

In the following embodiments of the present invention, the specific methods adopted in each step in the above general process will be further described.

Obtain the DOI response information of gamma photons incident on the crystal bar array

In one embodiment of the present invention, the GATE software based on Monte Cal simulation simulates the gamma photons incident on the crystal strip array at different angles, so as to obtain the DOI response information of this process.

The reason for the DOI effect is due to the penetration effect of the γ photon incident on the crystal strip at a certain non-right angle. When the γ photon is incident on the crystal strip vertically, as shown in Figure 5A, the γ photon will not penetrate to other crystal strips; but when the γ photon is incident on the crystal strip at a certain oblique angle, as shown in Figure 5B and Fig. Schematic diagram of 5C, the gamma photon penetrates the adjacent crystal strip along the direction of its propagation line, producing a blur effect. When the inclination angle of the gamma photon is different, the situation of penetrating the crystal strip is different, resulting in a different response, as shown in Figure 5B, the gamma photon passes through two crystal strips, and as shown in Figure 5C, it passes through three crystals strip. When the gamma photon interacts with the crystal strip, due to the same attenuation coefficient, it penetrates the same material at the same distance in the crystal strip, so the angle and position of the incident crystal strip array plane determine its in the crystal strip. Penetration, and when the side length of the crystal strip is short enough (such as 3.5mm), the incident position can be ignored, and the incident angle completely determines the penetration situation. Let the DOI response information we want be DOI, then DOI~ θ (θ is the incident angle of γ photons) is DOI(θ). According to the analysis results, as long as the size of the crystal strip is constant and the material is constant, no matter what the geometry of the detector is, the angular relationship between the γ photon and the crystal strip array is the same when it is incident. Available on detectors of any shape.

Specifically, this process can be simulated by GATE software to simulate the response of adjacent or even all crystal strips when gamma photons are incident on the array of crystal strips at different angles. This response includes position information and corresponding response information of these crystal strips. As shown in Figure 6, at the same time, there are N γ photons incident on the crystal strip numbered 1 at a certain angle θ ₁ , but only n ₁ attenuated and absorbed on crystal strip 1, and n ₂ attenuated absorbed on 2 On crystal strip No. 3, n ₃ attenuation absorbs on crystal strip No. 3, where n ₁ , n ₂ , and n ₃ are all smaller than N. If most of the γ photons respond in these crystal strips, that is, n ₁ +n ₂ +n ₃ >q*N, (here q is a large percentage window value that can be set, such as 99.5%, when n ₁ +n ₂ +n ₃ >q*N, it can be considered that the main effect is 1 , on crystal strips 2 and 3), then the positions of crystal strips 1, 2 and 3 and the response ratio information are the DOI response information of the crystal strip array we obtained, as shown in Table 1.

Table 1 Example of DOI response information

In addition, n ₁ , n ₂ , and n ₃ may be used directly for the DOI response ratios in the above DOI response information to replace the percentage values in the above table.

As mentioned above, regardless of the position of the crystal strip in the detector, the incident angles of the gamma photon and the crystal strip are the same, so the DOI response information of the gamma photon entering a crystal strip at different incident angles can be simulated, and then reused In other crystal strips you can. The range of incident angle θ is from (0°, 180°) (here means the open set of 0°-180°). Specifically, the start angle can take any small angle close to 0 degrees, and the cut-off angle can take any angle close to 180 degrees

degree to ensure that the simulated data can be used for more shapes of PET. The incident angle here is the angle between the incident ray and the edge of the crystal strip to the right of the incident point, the incident angle shown in Figure 7A is an acute angle, and the incident angle shown in Figure 7B is an obtuse angle. In addition, the method for determining the above-mentioned incident angle can also be determined through other relative coordinate systems, for example, the incident angle can be taken as the angle between the incident ray and the edge of the crystal strip to the left of the incident point, or the incident angle can be taken as the angle between the incident ray and the incident The angle between the normal to the plane of the crystal strip in the plane. The step size of the angle can take as many values as possible between the starting angle and the cut-off angle, such as 100 or 200 or even more, the present invention does not limit the number of angles taken, also in order to ensure that this simulation Can be used by more shapes of PET. From this, the step size can be obtained:

For the simulated crystal strip array, the data can also be organized in the form of Table 2, so as to facilitate the repeated use of the simulated results.

Table 2 DOI response information of γ photons incident on the crystal strip array

In Table 2, a represents the number of the incident crystal strip, a-x represents the xth crystal strip adjacent to the left of the incident crystal strip a, and similarly a+x represents the xth crystal strip adjacent to the right of the incident crystal strip a. θy represents different incident angles. p represents the simulated probability value, that is, the response ratio information, and its subscript represents the combination of different crystal numbers and incident angles. Table 2 only lists the four adjacent crystal strips on the left and right sides of the incident crystal strip. For more accurate calculation, more values can be taken to simulate more crystal strips, which is not limited in the present invention.

According to the analysis, as long as the size of the crystal strip is constant and the material is constant, no matter what the geometric shape of the detector is, the angular relationship between the γ photon and the crystal strip array is the same when it is incident, so when the γ photon is incident on the crystal strip array When simulating, it is not necessary to aim at a specific geometric shape of the detector, so that the present invention has stronger versatility.

The specific method flow of the Gate software Monte Carlo simulation that can be used is as follows:

Step 1: Build the detector model: set the properties of the crystal strip array, add a collimator in front of it, and set its size;

Step 2: Set the properties of the radioactive source: set the properties of the radioactive source and place it at a certain distance from the crystal strip array;

Step 3: Set up the physical process: set up the physical process of the reaction of the gamma photons emitted by the source with the array of crystal strips, and truly simulate the actual situation;

Step 4: Set data output;

Step 5: Constantly change the direction of the collimator to achieve the purpose of changing the incident angle of the γ photon.

According to another embodiment of the present invention, the above-mentioned method for obtaining DOI response information when gamma photons are incident on the array of crystal strips may also include using an experimental method to measure when gamma photons are incident on the array of crystal strips at different angles, in the corresponding one or Response conditions on multiple crystal strips, so as to obtain DOI response information of the corresponding one or more crystal strips when the γ photons are incident on the crystal strip array. The experimental method that can be used is similar to the above-mentioned specific simulation method, such as: place the tracer radioactive source used in the experiment at a certain distance from the detector array composed of crystal strips, place a collimator in front of the crystal strip array, and constantly change the collimator. The direction of the straightener is used to realize the different incident angles of the gamma photons, so that the responses of the gamma photons incident on the crystal strip at different incident angles can be obtained.

According to yet another embodiment of the present invention, the above-mentioned method for obtaining DOI response information when gamma photons are incident on the crystal strip array may also include analytically calculating the attenuation coefficient of the gamma photons in the crystal strips when the gamma photons are incident on the crystal strips at different angles When arraying, the response on the corresponding one or more crystal strips, so as to obtain the DOI response information of the corresponding one or more crystal strips when the γ photons are incident on the crystal strip array.

DOI response modes consistent with LOR determined by the timing of generating back-to-back gamma photon incident crystal strips type

As mentioned above, an event in the PET system involves a pair of crystal strips, so the DOI response information generated when the back-to-back γ photons are incident on the two crystal strips needs to conform to the response model of a response event, that is, the back-to-back γ photons incident on the crystal DOI response model for time-determined LOR.

The coincidence process plays a very important role in the establishment of the whole system response model. The crystal strips of detectors with different shapes are arranged in different ways and numbers are different, but no matter what the arrangement is, as long as the specifications of the crystal strips used are the same. The same method can be used to match the DOI response information of the crystal strip arrays at both ends of the back-to-back γ-photon incident event.

In one embodiment of the present invention, specifically, the obtained DOI response information of gamma photons incident on the crystal strip array is used to conform to a response model of an event. As shown in Figure 8, in order to reflect the generality, the crystal strips involved in the response event are numbered as a, b, c, d, e; the crystal strips at the other end are numbered f, g, h, i, j, k. The numbers here are for illustrative purposes only, and are not intended to limit the present invention. Assume that in the obtained DOI response information of γ-photons incident on the crystal strip array, the response probabilities (that is, the response ratio) of γ-photons incident on the crystal strip array corresponding to the incident angles θ ₁ and θ ₂ are shown in Table 3, where 0 means no Response, it can be seen from Table 3 that only the adjacent crystal strips on the left of the incident crystal strips b and g at both ends respond. It should be noted that the crystal numbers in Table 3 are relative numbers, which reflect the relative position information of each crystal strip, rather than the specific numbers in Figure 8. Therefore, when the incident angle is θ ₁ , a in Table 3 represents The incident crystal bar b in Fig. 8,

And when the incident angle is _θ2 , a in Table 3 represents the incident crystal strip g in Figure 8.

Table 3 DOI response information of γ photons incident on the crystal strip array at the incident angles of θ ₁ and θ ₂

Since these responses all occur at the same time, we use random combination to combine responding crystal strips in pairs to form the LOR determined by the responding crystal strip pair, and use the product of the response ratio in the corresponding DOI response information to represent the corresponding The annihilation event determined by the pair of crystal strips is the probability of the existence of LOR. We use Event(x, y) to represent the probability of the response produced by crystal strip x and crystal strip y, then we have

Event(x,y)=p _x ×p _y (1)

Among them, p _x and p _y represent the DOI response ratios of γ-photons incident on crystal strip x and γ-photon incident on crystal strip y, respectively.

In addition, if the response ratio of the acquired γ-photon incident into the DOI response information of the crystal strip array is a relative ratio, that is, the ratio between the attenuation and absorption numbers of different crystal strips, then the above probability product is the product of relative ratios.

In the incident situation shown in Figure 8, there are 2×3=6 kinds of response situations, and the response event probability is calculated according to formula (1). The event response model, that is, the DOI response model of the LOR determined by different crystal strip pairs is shown in Table 4, so that the system response of all events shown in FIG. 8 is obtained.

Table 4 fits the generated event DOI response model

response event response probability Event(b,g) p _a1 *p _a2 Event(b,h) p _a1 *p _a2-1 Event(b,i) p _a1 *p _a2-2 Event(c,g) p _a1-1 *p _a2 Event(c,h) p _a1-1 *p _a2-1 Event(c,i) p _a1-1 *p _a2-2

The response intensity of the matched DOI response model obtained here is a probability representation, which belongs to a relative proportional value, and only depends on the incident angle of the γ photon. Therefore, during the reconstruction process, different pixel points on the same LOR have the same DOI response probability after being matched. But in fact, due to the different positions of the pixels, the absolute intensity of the response to the same pair of crystal strips is different, that is, the N in the previous step is not the same, so in another embodiment of the present invention, The obtained DOI event probability model can also be further weighted according to the positions of different pixel points to obtain an absolute value DOI response model. The method for obtaining the weight value is as follows: the size of the solid angle opened by the incident crystal strip to the pixel point can be used to represent the size of the weight value.

The responses generated in Table 4 are based on crystal bar coordinates to organize the data, and this organized response can be used in the reconstruction algorithm based on List mode. If the subsequent iterative reconstruction algorithm is based on Sinogram, the above event response can also be transformed into the Sinogram data organization form, that is, the system response organization form based on the (r, θ) of the LOR line. The matching methods are the same. As long as the positions and incident angles of the two incident crystal strips are known, the desired system response can be obtained.

In an embodiment of the present invention, optionally, before the system response model is generated according to the DOI response model generated according to the coincidence, the symmetry information of the PET detector may also be obtained according to the geometric shape of the PET detector.

According to the above matching method, as long as the position and incident angle of the γ-photon incident crystal strip array are known, the response model of the event required for reconstruction can be obtained according to the obtained DOI response information obtained above, and then the system response model can be obtained. Therefore, in the process of obtaining the system response model for reconstruction, it becomes a key point to be able to determine the position information and incident angle information of the crystal strip array where the gamma photons are incident. Generally speaking, the responses of all used pixels can be calculated once according to the specific shape of the PET and the specific reconstruction algorithm (rebin+2D or 3D). However, there are special symmetries in a specific geometric structure, and by using these symmetries, the number of calculations can be simplified and the operation speed can be accelerated.

Specifically, in the static double flat panel detector, there are translational symmetry, axial symmetry and interchange symmetry, so only a few points need to be calculated to obtain the system response of all position points. In ring detectors, the system is generally designed in the shape of a regular polygon. Regular polygons have rotational symmetry and axis symmetry, and can also conform to the response of fewer points, so that the system response of all position points can be obtained. In addition, detectors of other shapes will have specific geometric characteristics, which can simplify the calculation.

Therefore, obtaining the symmetry information of the PET geometric shape can simplify the calculation times and speed up the operation speed.

Generating a System Response Model

In one embodiment of the present invention, a system response model is generated according to the obtained DOI response model generated according to the above.

Specifically, calculate and obtain the crystal strip position information and incident angle information corresponding to all pixels on the LOR, and generate a system response model based on the DOI response model obtained above, that is, the system response of all pixels used in the reconstruction algorithm information.

The system response model can be an array or matrix. Taking an array as an example, let the system response model be an array P[m][n], where m is used to uniquely determine a pixel point. For example, two-dimensional or three-dimensional coordinates can be used Determination; n is used to uniquely determine a LOR, for example, the numbers of the corresponding two crystal strips can be used, or the aforementioned (r, θ) organizational form; vice versa, that is, m is used to uniquely determine a LOR, and n is used for unique The embodiment of the present invention is not limited to determine a pixel; p[m][n] is an element of the array P[m][n], that is, the above-mentioned probability of LOR.

The method to calculate the system response model is to first determine a pixel point that needs to be calculated. Taking Fig. 8 as an example, the annihilation of this pixel point occurs, and the back-to-back γ photons are respectively incident on the crystal strip pair by θ ₁ and θ _2. The LOR is as follows: As shown in Figure 8, when m ₁ represents the pixel point, n ₁ , n ₂ , n ₃ , n ₄ , n ₅ and n ₆ represent (b, g), (b, h), (b, i) respectively, (c,g), (c,h), (c,i) formed LOR, then

p[m ₁ ][n ₁ ]=p _a1 *p _a2

p[m ₁ ][n ₂ ]=p _a1 *p _a2-1

p[m ₁ ][n ₃ ]=p _a1 *p _a2-2

p[m ₁ ][n ₄ ]=p _a1-1 *p _a2

p[m ₁ ][n ₅ ]=p _a1-1 *p _a2-1

p[m ₁ ][n ₆ ]=p _a1-1 *p _a2-2

After changing θ ₁ and θ ₂ , continue to calculate this pixel according to the above method; after that, continue to perform the above calculation on all pixels on the LOR according to the above method, and then the system response model can be obtained.

If in the step of generating the DOI response model of the LOR determined by the incident crystal strip pair, the probability product is also weighted, where p[m][n] is its weighted probability value.

In another embodiment of the present invention, if the symmetry information of the geometric shape of the PET detector has been obtained, the system response model can also be generated according to the obtained symmetry information and the DOI response model generated according to the above.

Specifically, according to the obtained symmetry information of the PET geometric shape, calculate and obtain the crystal strip position information and incident angle information corresponding to the pixel points on all LORs, and generate the system response model according to the obtained DOI response model generated by the above coincidence, namely System response information for all pixels.

This process can be generated in real time during the reconstruction process to save storage space; the generated system response can also be stored in the form of a matrix before the reconstruction process to speed up the calculation.

When generating the system response model here, there is one more step than previous methods to obtain the incident angle θ of the gamma photons, that is, the angle between the LOR line and the planes of the crystal strips at both ends, and the dimension of the system response matrix is also higher than that of the analytical method in the prior art many.

In the incident situation shown in Figure 8, five blurred system responses will be generated. As shown in Figure 9, the LOR lines that were originally solid lines, due to the DOI effect, produce the other five LOR lines shown as dotted lines, which are reflected in the generated system response through the above method, and then reflected in the reconstruction process.

It is worth noting that at the edge of the dual flat panel detector due to the truncation effect, the DOI effect will not be as complete as in the middle. As shown in Figure 10, when the gamma ray is incident on the crystal strip at the upper end, since there is no adjacent crystal strip at the left end of the incident crystal strip, a truncation effect occurs. Similarly, the edge of polygonally arranged blocks in the ring PET also has the same problem, as shown in Figure 11A, which is the real structure of the ring detector, and multiple blocks are spliced into a regular polygon structure. The crystal strips inside each block are distributed in an array, as shown in Figure 11B, where there is a gap at the edge of the block, there is an edge truncation like the flat PET. For the edge truncation processing of flat PET, since the truncated gamma photons hit the air when penetrating the edge crystal strip, it will not cause blurring, so it is only necessary to determine the response according to the actual edge situation, and the response ratio is based on the actual situation Just determine the percentage. In the situation shown in Figure 10, the response of the upper edge is only the incident crystal strip and one crystal strip, so the response ratio is set to 100%. The edge of the ring detector block is similar to the edge of the flat PET, and the same process is also performed.

image reconstruction

In an embodiment of the present invention, the obtained above-mentioned system response model is applied to an iterative reconstruction algorithm to perform image reconstruction. Commonly used iterative methods are MLEM based on probability model and its improved algorithm OSEM algorithm. In this embodiment, the OSEM algorithm is used as an example to illustrate the image reconstruction process based on the obtained system response model. In addition, the iterative reconstruction algorithm based on the obtained system response model can also be replaced by other iterative reconstruction methods that require the system response model as input, such as MLEM, ART, MLEM, OSEM, OSLS, MAP, etc.

Fig. 12 is a flow chart of the iterative reconstruction algorithm in the embodiment of the present invention. Every time the projection is calculated according to the new image F, the obtained system response model will be used to obtain the projection of the image F to the system, and then compared with the experimentally measured To correct the original image F by comparing with the projection of . In the iterative process of correction again and again, a more accurate image F will be obtained, that is, the image will be reconstructed. The specific steps are as follows:

Step S1201, setting an initial image F;

Step S1202, calculate the projection D of the image F to the system according to the obtained system response model;

Step S1203, compare with the projection D' measured by experiment;

Step S1204, calculating the correction coefficient and updating the image F;

Step S1205, judging whether the stop rule is satisfied, if not, re-execute step S1202, and input the updated image F into step S1202; otherwise, if satisfied, execute step S1206;

Step S1206, terminating the iterative process.

The obtained system response model can be calculated in real time during the process of iterative reconstruction algorithm, or it can be pre-calculated and stored on the corresponding device, and can be directly read and used during reconstruction.

It should be noted that any stopping rule in the prior art that is suitable for the algorithm used may be used as the stopping rule, and details will not be described here.

In addition, the above iterative algorithm steps are only used to illustrate the process of image reconstruction based on the above obtained system response model, the embodiment of the present invention is not limited thereto, the system response model obtained in the embodiment of the present invention can be applied to any One in an iterative algorithm for image reconstruction that requires a system response model as input. The method and device for obtaining the PET system response model and image reconstruction provided by the present invention, by obtaining the DOI response information of the gamma photon incident on the crystal strip array, conforms to the DOI response model of the LOR determined by the back-to-back gamma photon incident crystal strip pair, and will It is applied to the reconstruction process to solve the image blur problem caused by the DOI effect. While obtaining high-quality images, it has good portability and can save time and storage costs, as long as the specifications of the crystal strips used in PET are the same. , no matter how the geometry of the PET changes, it is not necessary to repeatedly simulate the DOI response information generated by the γ-photon incident crystal strip, and the process method is simple, and the system response model can be generated in real time during the reconstruction process, which saves time costs and storage costs.

Fig. 13 is a schematic diagram of an apparatus for acquiring a PET system response model and image reconstruction disclosed by an embodiment of the present invention. As shown in Figure 13, the device specifically includes:

An acquisition module 1301, configured to acquire DOI response information of gamma photons incident on the crystal strip array;

Specifically, the method for obtaining the DOI response information of gamma photons incident on the array of crystal strips includes: using Monte Carlo to simulate the response of one or more crystal strips when gamma photons are incident on the array of crystal strips at different angles situation, so as to obtain the DOI response information of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array; when the gamma photon is incident on the crystal strip array at different angles, the corresponding one or more The response situation on the crystal strips, so as to obtain the DOI response information of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array; When the angle is incident on the crystal strip array, the response on the corresponding one or more crystal strips, so as to obtain the DOI response information of the corresponding one or more crystal strips when the γ photon is incident on the crystal strip array.

The matching module 1302 is used for generating a DOI response model of LOR determined by back-to-back gamma photon incident crystal bar array according to the obtained DOI response information of the gamma photon incident crystal bar array;

The response of a pair of crystal strips involved in a response event is matched to obtain the DOI response model of the crystal strip to the determined LOR. A specific matching method may be a probabilistic matching method.

Calculation module 1303, configured to calculate and generate a system response model according to the DOI response model obtained and generated;

Specifically, calculate and obtain the crystal strip position information and incident angle information corresponding to all pixels on the LOR, and generate a system response model based on the DOI response model obtained above, that is, the system response of all pixels used in the reconstruction algorithm information.

This process can be generated in real time during the reconstruction process to save storage space; the generated system response can also be stored in the form of a matrix before the reconstruction process to speed up the calculation.

A reconstruction module 1304, configured to perform image reconstruction according to the acquired system response model.

According to the obtained system response model, the image reconstruction process is carried out through an iterative algorithm.

The method and device for obtaining the PET system response model and image reconstruction provided by the present invention, by obtaining the DOI response information of the gamma photon incident on the crystal strip array, conforms to the DOI response model of the LOR determined by the back-to-back gamma photon incident crystal strip pair, and will It is applied to the reconstruction process to solve the image blur problem caused by the DOI effect. While obtaining high-quality images, it has good portability and can save time and storage costs, as long as the specifications of the crystal strips used in PET are the same. , no matter how the geometry of the PET changes, it is not necessary to repeatedly simulate the DOI response information generated by the γ-photon incident crystal strip, and the process method is simple, and the system response model can be generated in real time during the reconstruction process, which saves time costs and storage costs.

Exemplary embodiments of the present invention have been specifically shown and described above. It should be understood that the invention is not limited to the disclosed embodiments, but on the contrary, it is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims (26)

1. A method of obtaining a system response model for positron emission tomography, comprising: Obtain the depth effect response information of the gamma photon incident on the crystal strip array; According to the depth effect response information, the depth effect response model that generates the response line determined by the back-to-back gamma photon incident crystal strip pair time; and, A system response model is generated based on the depth effect response model. 2. The method for obtaining the system response model of positron emission tomography according to claim 1, wherein the method for obtaining the depth effect response information of gamma photons incident on the crystal strip array comprises: Simulate the response of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array at different angles by software, so as to obtain the response of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array Depth effect response information. 3. The method for obtaining the system response model of positron emission tomography according to claim 1, wherein the method for obtaining the depth effect response information of gamma photons incident on the crystal strip array comprises: Using an experimental method to measure the response of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array at different angles, so as to obtain the gamma photon incident on the crystal strip array of the corresponding one or more crystal strips The depth effect response information. 4. The method for obtaining the system response model of positron emission tomography according to claim 1, wherein the method for obtaining the depth effect response information of gamma photons incident on the crystal strip array comprises: Using the attenuation coefficient of the gamma photon in the crystal strip to analyze and calculate the response of the gamma photon on the corresponding one or more crystal strips when it is incident on the crystal strip array at different angles, so as to obtain the corresponding one or more crystal strips Depth-effect response information of gamma photons incident on crystal bar arrays. 5. The method for obtaining a system response model of positron emission tomography according to any one of claims 1 to 4, wherein the depth effect response information of the gamma photon incident on the crystal bar array includes: crystal bar number , the depth effect scale and the angle of incidence. 6. The method for obtaining a system response model of positron emission tomography according to claim 5, wherein the depth effect ratio is the number of gamma photons incident on a crystal strip with an incident angle to the incident angle with the incident angle The ratio of the number of all gamma photons, or the number of gamma photons incident on a crystal strip at an incident angle. 7. The method for obtaining a system response model of positron emission tomography according to claim 5, wherein the method for determining the incident angle comprises: taking the angle between the incident ray and the edge of the crystal strip to the right or left of the incident point angle, or take the angle between the incident ray and the normal of the crystal strip plane in the incident plane. 8. The method for obtaining the system response model of positron emission tomography according to claim 5, wherein the response information according to the depth effect conforms to the depth of the response line determined when generating back-to-back gamma photon incident crystal strip pairs Methods for effect-response modeling include: pairwise combining responsive crystal strips to form a response line determined by responsive crystal strip pairs; and, Multiplying the depth effect ratios in the respective depth effect response information of the pair of crystal strips to obtain the depth effect response model. 9. The method for obtaining the system response model of positron emission tomography according to claim 5, wherein the response information according to the depth effect conforms to the depth of the response line determined when generating back-to-back gamma photon incident crystal strip pairs Methods for effect-response modeling include: Combining the responding crystal strips in pairs to form a response line determined by the responding crystal strip pair; multiplying the depth effect ratios in the respective depth effect response information of the crystal strip pairs; and, The product of the above steps is further weighted according to the position information of different pixels, wherein the weight is determined by the solid angle opened by the incident crystal strip to the pixel, so as to obtain the depth effect response model. 10. The method for obtaining the system response model of positron emission tomography according to claim 1, wherein the data organization form of the depth effect response model of the crystal strip pair comprises: list mode (List mode) form and sinogram (Sinogram) form. 11. The method for obtaining a system response model of positron emission tomography according to claim 1, wherein, before generating the system response model according to the depth effect response model, further comprising The geometric shape of the detector obtains its symmetry information, and the generating the system response model according to the depth effect response model is further to generate the system response model according to the obtained symmetry information and the depth effect response model. 12. The method for obtaining the system response model of positron emission tomography according to claim 1, wherein the method for generating the system response model according to the depth effect response model comprises: calculating the pixel points on all response lines system response information, thereby generating the system response model. 13. A method of image reconstruction for positron emission tomography, comprising: Obtain the depth effect response information of the gamma photon incident on the crystal strip array; According to the depth effect response information, it conforms to the depth effect response model for generating the response lines determined by the back-to-back gamma photon incident crystal strips; generating a system response model based on the depth effect response model; and, Image reconstruction is performed according to the system response model described. 14. The image reconstruction method for obtaining positron emission tomography according to claim 13, wherein the method for obtaining the depth effect response information of gamma photons incident on the crystal strip array comprises: Simulate the response of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array at different angles by software, so as to obtain the response of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array Depth effect response information. 15. The image reconstruction method for obtaining positron emission tomography according to claim 13, wherein the method for obtaining the depth effect response information of gamma photons incident on the crystal strip array comprises: Using an experimental method to measure the response of the corresponding one or more crystal strips when the gamma photons are incident on the crystal strip array at different angles, so as to obtain the gamma photon incident on the crystal strip array of the corresponding one or more crystal strips The depth effect response information. 16. The image reconstruction method for obtaining positron emission tomography according to claim 13, wherein the method for obtaining the depth effect response information of gamma photons incident on the crystal strip array comprises: Using the attenuation coefficient of the gamma photon in the crystal strip to analyze and calculate the response of the gamma photon on the corresponding one or more crystal strips when it is incident on the crystal strip array at different angles, so as to obtain the corresponding one or more crystal strips Depth-effect response information of gamma photons incident on crystal bar arrays. 17. The image reconstruction method for obtaining positron emission tomography according to any one of claims 13 to 16, wherein the depth effect response information of the gamma photons incident on the crystal strip array includes: crystal strip number, depth Effect scale and angle of incidence. 18. The image reconstruction method for obtaining positron emission tomography according to claim 17, wherein the depth effect ratio is the number of gamma photons incident on a crystal strip at an incident angle to all the gamma photons incident at the incident angle The ratio of the number of gamma photons, or the number of gamma photons incident on a crystal strip at an incident angle. 19. The image reconstruction method for obtaining positron emission tomography according to claim 17, wherein the method for determining the incident angle comprises: taking the angle between the incident ray and the edge of the crystal strip to the right or left of the incident point, Or take the angle between the incident ray and the normal to the crystal strip plane in the incident plane. 20. The image reconstruction method for obtaining positron emission tomography according to claim 17, wherein, the depth effect response according to the response line determined when generating back-to-back gamma photon incident crystal strip pairs according to the depth effect response information Model methods include: pairwise combining responsive crystal strips to form a response line determined by responsive crystal strip pairs; and, Multiplying the depth effect ratios in the respective depth effect response information of the pair of crystal strips to obtain the depth effect response model. 21. The image reconstruction method for obtaining positron emission tomography according to claim 17, wherein the depth effect response according to the response line determined when generating back-to-back gamma photon incident crystal strip pairs according to the depth effect response information Model methods include: Combining the responding crystal strips in pairs to form a response line determined by the responding crystal strip pair; multiplying the depth effect ratios in the respective depth effect response information of the crystal strip pairs; and, The product of the above steps is further weighted according to the position information of different pixels, wherein the weight is determined by the solid angle opened by the incident crystal strip to the pixel, so as to obtain the depth effect response model. 22. The image reconstruction method of obtaining positron emission tomography according to claim 13, wherein, the data organization form of the depth effect response model conforming to the obtained LOR comprises: list mode (List mode) form and sinogram ( Sinogram) form. 23. The image reconstruction method for obtaining positron emission tomography according to claim 13, wherein, before said generating a system response model according to said depth effect response model, further comprising detection according to said positron emission tomography The geometric shape of the sensor is used to obtain its symmetry information, and the generating the system response model according to the depth effect response model is further to generate the system response model according to the obtained symmetry information and the depth effect response model. 24. The image reconstruction method for acquiring positron emission tomography according to claim 13, wherein the method for generating a system response model according to the depth effect response model comprises: calculating system responses of pixels on all response lines information to generate the system response model. 25. The image reconstruction method for obtaining positron emission tomography according to claim 13, wherein the method for performing image reconstruction according to the system response model comprises: set an image F; calculating the projection D of the image F to the system according to the system response model; Compare with the projection D' measured in the experiment; calculating correction coefficients and updating said image F; and, Judging whether the stop rule is satisfied: if not, re-execute the calculation of the projection D of the image F to the system according to the system response model, wherein the image F is an updated image F; if it is satisfied, terminate Iteration process. 26. A device for obtaining a system response model and image reconstruction for positron emission tomography, comprising: An acquisition module, configured to acquire depth effect response information of gamma photons incident on the crystal strip array; The coincidence module is used to conform to generate a depth effect response model of a response line determined by back-to-back gamma photon incident crystal strips according to the acquired depth effect response information; a calculation module, configured to calculate and generate a system response model according to the depth effect response model; and, The reconstruction module is used for performing image reconstruction according to the system response model.