CN103996213A - PET image rebuilding method and system - Google Patents

Abstract

The invention provides a PET (Positron Emission Tomography) image rebuilding method and a PET image rebuilding system. The method comprises the following steps that calibration point data of a PET image is obtained; an image rebuilding model based on the maximum likelihood-expectation maximization algorithm is built according to the statistical characteristics of the calibration point data; a point spread function is utilized for correcting the image rebuilding model, and a corrected image rebuilding model is obtained; actually measured data is input into the corrected image rebuilding model, and the PET rebuilding image is obtained. The PET image rebuilding method and the PET image rebuilding method system have the advantages that the image rebuilding is carried out on the basis of a sonogram space, the algorithm complexity is reduced from the traditional O (N6) to O(N4), the rebuilding operation time is reduced, and in addition, the symmetry of the sonogram space is utilized, so experiment calibration points are greatly reduced, and the requirements on storage and calculation resources can be reduced.

Description

A kind of PET image rebuilding method and system

Technical field

The present invention relates to the technical field of image processing of medical image, particularly relate to a kind of PET image rebuilding method and system.

Background technology

Positron emission tomography (Positron Emission Computed Tomography, PET) as a kind of instrument of non-intervention quantitative examination live body function activity, be applied to just more and more widely clinical diagnosis, especially the early diagnosis of disease.PET spatial resolution is one of core index of PET system, the certain physical characteristics of positron and the limitation of data collecting mechanism, cause spatial resolution inevitably to reduce, show on image for image blurring, this has restricted PET system and has required the application in high-resolution field at some, as the discovery of infantile tumour, cerebral function imaging etc.

Fig. 1 shows the geometry schematic diagram of PET gauging ring, and wherein, A point is the position at radioactive point source place, and gauging ring is arranged according to regular polygon by a plurality of detector module T, and each detector module T inside comprises again many crystal detections.When a positron, at A point, bury in oblivion, produce a pair of reverse gamma-ray photon and get to respectively on corresponding detector module, as the solid line LOR1 of ordering through A in Fig. 1.Because gamma-ray photon has certain kinetic energy, in crystal detection, through a segment distance, just can be caught in, this segment distance is called depth of interaction (Depth of Interact, DOI).The crystal that captures gamma-ray photon can be launched faint visible ray, after the photomultiplier amplification by rear end, converts electric signal to.The line that captures a pair of crystal of gamma-ray photon is called line response (Line of Response, LOR), is used for the position that mark buries in oblivion.When a pair of reverse gamma-ray photon is mapped to detector module along the solid line LOR1 shown in Fig. 1 is oblique respectively, impact due to DOI, making the position of the LOR that detects is not real incoming position, but the corresponding position of dotted line LOR2 in Fig. 1, thereby has affected the spatial resolution of PET.

If make the complete normal incidence of gamma-ray photon to detector module, as shown in Figure 2, a and b are detector module, S (a) and S (b) are respectively the response of detector module a and b, in crystal counter, may there is the scattering of certain probability in gamma-ray photon, scattering will change the working direction of gamma-ray photon, and the signal therefore detecting on detector a and b, by the statistical distribution presenting centered by incidence point, can affect the spatial resolution of PET equally.Other some factors, as positron moving range (Positron Range), partial volume effect (Partial Volume Effect, PVE), non-colinear (Non-Colinearity) etc. also can have influence on the spatial resolution of PET.

In order to improve the spatial resolution of PET, prior art adopts the method for image reconstruction, for example first adopt maximum likelihood expectation maximization algorithm (Maximum-Likelihood Expectation-Maximization, ML-EM) image is carried out to interative computation, by spatial resolution, recover to be optimized again, still, the method adopting when current spatial resolution is recovered is to carry out at image space, computation complexity is higher, is O (N ⁶), in addition, need high accuracy three-dimensional space positioning apparatus to carry out data acquisition, and need to gather a large amount of data of calibration point (approximately thousands of), store, process these data and need to expend a large amount of storages and computational resource.

Summary of the invention

The shortcoming of prior art in view of the above, the object of the present invention is to provide a kind of PET image rebuilding method and system, more for solving prior art data acquisition request data of calibration point higher, that need to gather, and spatial resolution higher problem of computation complexity while recovering.

For achieving the above object and other relevant objects, the invention provides a kind of PET image rebuilding method, described PET image rebuilding method at least comprises:

Obtain the data of calibration point of PET image;

According to the statistical nature of described data of calibration point, build the image reconstruction model based on maximum likelihood expectation maximization algorithm;

Utilize point spread function to revise described image reconstruction model, obtain revised image reconstruction model;

Measured data is inputted to described revised image reconstruction model, obtain PET and rebuild image.

Preferably, the data of calibration point of obtaining PET image further comprises:

Build virtual pet system environments and be contained in the virtual point source in described virtual pet system environments;

In described virtual pet system environments, scan described virtual point source, obtain the position data of calibration point;

According to the position data of described calibration point, generate the sinogram of described calibration point.

Preferably, described calibration point is arranged in 1/2 surface area by the central point of PET gauging ring and the formed sector region of single detector module of corresponding described central point, wherein, described PET gauging ring comprises a plurality of detector modules of arranging according to regular polygon.

Preferably, after the sinogram that generates described calibration point according to the position data of described calibration point, also comprise: the sinogram of described calibration point is carried out to interpolation arithmetic, to proofread and correct described sinogram.

Preferably, the image reconstruction model building based on maximum likelihood expectation maximization algorithm further comprises:

Obtain a default image;

Calculate described default image along the projection of each φ direction, obtain estimating sinogram, wherein, φ is through the light of radioactive point source and the angle of horizontal direction in PET system;

Relatively estimate the error between sinogram and actual measurement sinogram, obtain error sinogram;

Described error sinogram is carried out to back projection, to upgrade estimated image, the row iteration computing of going forward side by side.

Preferably, the formula of described interative computation is:

${\mathrm{\λ}}_i^{k+1}=\frac{{\mathrm{\λ}}_i^k}{{\mathrm{\Σ}}_{m=1}^M{F}_{\mathrm{mi}}{A}_m}\×\underset{j=1}{\overset{M}{\mathrm{\Σ}}}{F}_{\mathrm{ji}}\left(\frac{A_j{p}_j}{A_j{\mathrm{\Σ}}_{l=1}^N{F}_{\mathrm{jl}}{\mathrm{\λ}}_l^k+R_j+S_j}\right)$

Wherein, the value of i pixel of presentation video after the k time iteration, p _jthe collection value that represents j data point in sinogram, A, R, S represent respectively attenuation correction coefficient, random coincidence event number, scattering events number, F _jithe probability that the event that expression produces from i pixel collects in j data point of sinogram, M represents the number of sinogram spatial data points, the number of N presentation video pixel.

Preferably, utilize point spread function to revise described image reconstruction model, obtain revised image reconstruction model and further comprise:

Set up PSF model, described PSF model is:

$\mathrm{PSF}=\exp (-\frac{{(\mathrm{xr}-x_0)}^2}{{2\mathrm{\&sigma;}}^2}),$ $\mathrm{\&sigma;}=\left\{\begin{array}{cc}{\mathrm{\&sigma;}}_{\mathrm{left}}& \mathrm{xr}<x_0\\ {\mathrm{\&sigma;}}_{\mathrm{right}}& \mathrm{xr}>=x_0\end{array}\right.;$ Wherein, X _rfor departing from centre distance, x ₀x corresponding to peak position estimating _rposition, σ _leftand σ _rightbe respectively x ₀the distribution variance on the left side and the right;

By described PSF model substitution image reconstruction model, obtain revised image reconstruction model.

Preferably, by described PSF model substitution image reconstruction model, obtain revised image reconstruction model and further comprise:

The orthogonal projection transforming function transformation function of definition image reconstruction model is P () _j=∑ _if _ji(), back project function is BP () _i=∑ _jf _ji(), utilizes described PSF model to align projective transformation function and back project function is revised, and obtains revised orthogonal projection transforming function transformation function back project function $\mathrm{BP}{(\&CenterDot;)}_i^{\mathrm{PSF}}={\mathrm{\&Sigma;}}_j\left[F_{\mathrm{ji}}(\&CenterDot;*{\mathrm{PSF}}^T)\right];$

By described revised orthogonal projection transforming function transformation function and back project function substitution image reconstruction model, obtain revised image reconstruction model.

Correspondingly, the present invention also provides a kind of PET image re-construction system, and described PET image re-construction system at least comprises:

Data of calibration point acquisition device, for obtaining the data of calibration point of PET image;

Image reconstruction model construction device, for according to the statistical nature of described data of calibration point, builds the image reconstruction model based on maximum likelihood expectation maximization algorithm;

Image reconstruction model correcting device, for utilizing point spread function to revise described image reconstruction model, obtains revised image reconstruction model;

PET rebuilds image calculation device, for measured data being inputted to described revised image reconstruction model, obtains PET and rebuilds image.

Preferably, described data of calibration point acquisition device comprises:

System environments virtual module, for the virtual point source that builds virtual pet system environments and be contained in described virtual pet system environments;

Data of calibration point analog module, for scanning described virtual point source in described virtual pet system environments, obtains the position data of calibration point;

Sinogram generation module, for generating the sinogram of described calibration point according to the position data of described calibration point.

Preferably, described calibration point is arranged in 1/2 surface area by the central point of PET gauging ring and the formed sector region of single detector module of corresponding described central point, wherein, described PET gauging ring comprises a plurality of detector modules of arranging according to regular polygon.

Preferably, also comprise sinogram correction module, for the sinogram of described calibration point is carried out to interpolation arithmetic, to proofread and correct described sinogram.

Preferably, described image reconstruction model construction device comprises:

Default image collection module, for obtaining a default image;

Estimate sinogram computing module, for calculating described default image along the projection of each φ direction, obtain estimating sinogram, wherein, φ is through the light of radioactive point source and the angle of horizontal direction in PET system;

Comparison module, for relatively estimating the error between sinogram and actual measurement sinogram, obtains error sinogram;

Interative computation module, for described error sinogram is carried out to back projection, to upgrade estimated image, the row iteration computing of going forward side by side.

Preferably, the formula of described interative computation is:

${\mathrm{\&lambda;}}_i^{k+1}=\frac{{\mathrm{\&lambda;}}_i^k}{{\mathrm{\&Sigma;}}_{m=1}^M{F}_{\mathrm{mi}}{A}_m}\&times;\underset{j=1}{\overset{M}{\mathrm{\&Sigma;}}}{F}_{\mathrm{ji}}\left(\frac{A_j{p}_j}{A_j{\mathrm{\&Sigma;}}_{l=1}^N{F}_{\mathrm{jl}}{\mathrm{\&lambda;}}_l^k+R_j+S_j}\right)$

Wherein, the value of i pixel of presentation video after the k time iteration, p _jthe collection value that represents j data point in sinogram, A, R, S represent respectively attenuation correction coefficient, random coincidence event number, scattering events number, F _jithe probability that the event that expression produces from i pixel collects in j data point of sinogram, M represents the number of sinogram spatial data points, the number of N presentation video pixel.

Preferably, described image reconstruction model correcting device comprises:

PSF model building module, for setting up PSF model, described PSF model is:

$\mathrm{PSF}=\exp (-\frac{{(\mathrm{xr}-x_0)}^2}{{2\mathrm{\&sigma;}}^2}),$ $\mathrm{\&sigma;}=\left\{\begin{array}{cc}{\mathrm{\&sigma;}}_{\mathrm{left}}& \mathrm{xr}<x_0\\ {\mathrm{\&sigma;}}_{\mathrm{right}}& \mathrm{xr}>=x_0\end{array}\right.;$ Wherein, X _rfor departing from centre distance, x ₀x corresponding to peak position estimating _rposition, σ _leftand σ _rightbe respectively x ₀the distribution variance on the left side and the right;

PSF correcting module, for by described PSF model substitution image reconstruction model, obtains revised image reconstruction model.

As mentioned above, a kind of PET image rebuilding method of the present invention and system, have following beneficial effect:

First, the present invention adopts maximum likelihood expectation maximization algorithm and the correction of PSF model is combined, and the process of image reconstruction is carried out in sinogram (Sinogram) space, and algorithm complex is by traditional O (N ⁶) be reduced to O (N ⁴), reduced reconstruction operation time.

Secondly, utilize the symmetry in sinogram space, make the calibration point of experiment only need to cover 1/2 surface area by the central point of PET gauging ring and the formed sector region of single detector module of corresponding described central point, thereby the quantity of calibration point is greatly reduced, reduced the demand to storage, computational resource.

Again, the method by Monte Carlo (Monte-Carlo) simulation obtains data of calibration point, without relying on positioning mechanical system, has avoided the error causing due to a Source size and mechanical positioning precision.

Accompanying drawing explanation

Fig. 1 is shown as the geometry schematic diagram of PET gauging ring of the prior art.

Fig. 2 is shown as the path schematic diagram of the gamma-ray photon normal incidence of PET gauging ring of the prior art.

Fig. 3 is shown as the mapping relations schematic diagram of calibration point of the present invention and its sinogram.

Fig. 4 is shown as the band of position schematic diagram of calibration point of the present invention.

Fig. 5 is shown as the mapping relations schematic diagram of calibration point and its sinogram in Fig. 4.

Fig. 6 is shown as the schematic diagram of sinogram interpolation arithmetic of the present invention.

Fig. 7 is shown as the process flow diagram of PET image rebuilding method of the present invention.

Fig. 8 is shown as the embodiment process flow diagram of PET image rebuilding method of the present invention.

Fig. 9 is shown as the schematic diagram of PET equipment for reconstructing image of the present invention.

Element numbers explanation

T detector module

DOI depth of interaction

LOR1 True Data

LOR2 measured data

A, b detector module

S (a) detector module a response

S (b) detector module b response

O gauging ring central point

F calibration point region

R gauging ring radius

X _rdepart from centre distance

φ is through the solid line of radioactive point source and the angle of horizontal direction

100 PET image re-construction systems

101 data of calibration point acquisition device

102 image reconstruction model construction devices

103 image reconstruction model correcting devices

104 PET rebuild image calculation device

1011 system environments virtual modules

1012 data of calibration point analog modules

1013 sinogram generation modules

1014 sinogram correction modules

1021 default image collection module

1022 estimate sinogram computing module

1023 comparison modules

1024 interative computation modules

1031 PSF model building modules

1032 PSF correcting modules

S1～S4 step

S11～S16 step

Embodiment

Below, by specific instantiation explanation embodiments of the present invention, those skilled in the art can understand other advantages of the present invention and effect easily by the disclosed content of this instructions.The present invention can also be implemented or be applied by other different embodiment, and the every details in this instructions also can be based on different viewpoints and application, carries out various modifications or change not deviating under spirit of the present invention.

Refer to the process flow diagram of Fig. 7 PET image rebuilding method of the present invention.

It should be noted that, the diagram providing in the present embodiment only illustrates basic conception of the present invention in a schematic way, satisfy and only show with assembly relevant in the present invention in graphic but not component count, shape and size drafting while implementing according to reality, during its actual enforcement, kenel, quantity and the ratio of each assembly can be a kind of random change, and its assembly layout kenel also may be more complicated.

Described PET image rebuilding method at least comprises:

Step S1: the data of calibration point of obtaining PET image;

Step S2: according to the statistical nature of described data of calibration point, build the image reconstruction model based on maximum likelihood expectation maximization algorithm;

Step S3: utilize point spread function to revise described image reconstruction model, obtain revised image reconstruction model;

Step S4: measured data is inputted to described revised image reconstruction model, obtain PET and rebuild image.

It should be noted that, obtain the method for the data of calibration point of PET image and can obtain the data of calibration point of PET image according to Monte Carlo (Monte Carlo) analogy method, specifically comprise:

Build virtual pet system environments and be contained in the virtual point source in described virtual pet system environments;

In described virtual pet system environments, scan described virtual point source, obtain the position data of calibration point;

According to the position data of described calibration point, generate the sinogram of described calibration point.

It should be noted that, described virtual pet system environments can build by writing a set of computer software based on Monte Carlo simulation.The principle of Monte Carlo simulation is known, and its accuracy and reliability are also confirmed by many scientific research documents.The geometry of specific PET gauging ring and physical parameter are inputted to this computer software, just built a virtual pet system environments.Described geometry and physical parameter comprise that physical dimension, locus, basic material and the judgement of detector module meet the elementary tactics of event etc.Under acceptable trueness error, virtual detection data is consistent with the data that actual PET system is surveyed, thereby can be for design of graphics as reconstruction model.

Described virtual point source can be designed to have the geometric point of infinitely small bulk, and this point has certain radioactivity, and infinitesimal virtual point can effectively be eliminated the error of bringing due to the size of point source.

Can also obtain by imaging device the data of calibration point of PET image, particularly, by imaging device, gather the detection data before PET imaging, obtain correction parameter values and the system matrix of imaging device simultaneously, and by imaging device, obtained detection data is carried out to Data correction, detection data after being proofreaied and correct, the detection data of usining after proofreading and correct is as the PET data as reconstruction model for design of graphics.

Detection data before described PET imaging refers to the line of a pair of crystal that captures gamma-ray photon, is called line response (Line of Response, LOR).By the LOR collecting according to off-centered distance X _rthe slanted angle φ of (central point of PET gauging ring is to the vertical range of the single detector module of corresponding described central point) and horizontal direction (through the solid line of radioactive point source and the angle of horizontal direction) tissue is arranged and is obtained with the steric image data of sinogram (Sinogram).The corresponding a line X of angle of inclination φ of identical on sinogram (or very approaching) _rdata are called a projection (Projection), and this is projected as a sinogram.As Fig. 3 has provided the relation of a data for projection and sinogram, the central point that in figure, stain O is gauging ring, 1,2,3 and 4 four LOR of an angle of inclination φ correspondence of obtaining with this central point, this four LOR position in sinogram is Fig. 3 the right X _rround dot mark 1,2,3 and 4 in-φ coordinate system.

The present invention preferably adopts Monte-carlo Simulation Method to obtain the data of calibration point of PET image.Those skilled in the art can also adopt method for distinguishing to obtain the data of calibration point of PET image, and the present invention only for illustrating, is not for limiting the application at this.

The described image reconstruction model based on maximum likelihood expectation maximization algorithm (Maximum-Likelihood Expectation-Maximization, ML-EM) further comprises:

Obtain a default image;

Calculate described default image along the projection of each φ direction, obtain estimating sinogram, wherein, φ is through the light of radioactive point source and the angle of horizontal direction in PET system;

Relatively estimate the error between sinogram and actual measurement sinogram, obtain error sinogram;

Described error sinogram is carried out to back projection, to upgrade estimated image, the row iteration computing of going forward side by side.

Wherein, the formula of described interative computation is:

${\mathrm{\&lambda;}}_i^{k+1}=\frac{{\mathrm{\&lambda;}}_i^k}{{\mathrm{\&Sigma;}}_{m=1}^M{F}_{\mathrm{mi}}{A}_m}\&times;\underset{j=1}{\overset{M}{\mathrm{\&Sigma;}}}{F}_{\mathrm{ji}}\left(\frac{A_j{p}_j}{A_j{\mathrm{\&Sigma;}}_{l=1}^N{F}_{\mathrm{jl}}{\mathrm{\&lambda;}}_l^k+R_j+S_j}\right)$ (formula 1)

Wherein, the value of i pixel of presentation video after the k time iteration, p _jthe collection value that represents j data point in sinogram, A, R, S represent respectively attenuation correction coefficient, random coincidence event number, scattering events number, F _jithe probability that the event that expression produces from i pixel collects in j data point of sinogram, M represents the number of sinogram spatial data points, the number of N presentation video pixel.

Described default image is the arbitrary image of a hypothesis, and the gray-scale value that conventionally defines the pixel in described image is 1.

Utilize point spread function to revise described image reconstruction model, obtain revised image reconstruction model and further comprise:

Set up PSF model, described PSF model is:

$\mathrm{PSF}=\exp (-\frac{{(\mathrm{xr}-x_0)}^2}{{2\mathrm{\&sigma;}}^2}),$ $\mathrm{\&sigma;}=\left\{\begin{array}{cc}{\mathrm{\&sigma;}}_{\mathrm{left}}& \mathrm{xr}<x_0\\ {\mathrm{\&sigma;}}_{\mathrm{right}}& \mathrm{xr}>=x_0\end{array}\right.$

(formula 2)

Wherein, X _rfor departing from centre distance, x ₀x corresponding to peak position estimating _rposition, σ _leftand σ _rightbe respectively x ₀the distribution variance on the left side and the right;

By described PSF model substitution image reconstruction model, obtain revised image reconstruction model.

It should be noted that x ₀, σ _leftand σ _rightbe according to formula 2, each data for projection of calibration point carried out to matching obtains.

Utilize described PSF model to align projective transformation function and back project function is revised, by revised orthogonal projection transforming function transformation function and back project function substitution image reconstruction model, obtain revised image reconstruction model.

The orthogonal projection transforming function transformation function that particularly, can define image reconstruction model is P () _j=∑ _if _ji(), back project function is BP () _i=∑ _jf _ji(), utilizes described PSF model to align projective transformation function and back project function is revised, and obtains revised orthogonal projection transforming function transformation function back project function $\mathrm{BP}{(\&CenterDot;)}_i^{\mathrm{PSF}}={\mathrm{\&Sigma;}}_j\left[F_{\mathrm{ji}}(\&CenterDot;*{\mathrm{PSF}}^T)\right];$

By described revised orthogonal projection transforming function transformation function and back project function substitution image reconstruction model, obtain revised image reconstruction model.

It should be noted that, point spread function (point Spread Function, PSF), is called again explorer response, and in sinogram space, different projection angle φ correspondences different eccentric distance X _r, and different eccentric distance X _rcorresponding different PSF.When image reconstruction model is formula 1, by orthogonal projection transforming function transformation function P () _j=∑ _if _ji() and back project function BP () _i=∑ _jf _ji() substitution formula 1 is simplified and is obtained:

${\mathrm{\&lambda;}}_i^{k+1}=\frac{{\mathrm{\&lambda;}}_i^k}{\mathrm{PB}{\left(A\right)}_i}\&times;\mathrm{BP}{\left(\frac{A\&times;p}{A\&times;P\left({\mathrm{\&lambda;}}^k\right)+R+S}\right)}_i$

(formula 3)

By revised orthogonal projection transforming function transformation function and back project function substitution formula 3, replace original orthogonal projection back project, obtain the ML-EM formula that carrying space resolution is recovered:

${\mathrm{\&lambda;}}_i^{k+1}=\frac{{\mathrm{\&lambda;}}_i^k}{\mathrm{PB}{\left(A\right)}_i^{\mathrm{PSF}}}\&times;\mathrm{BP}{\left(\frac{A\&times;p}{A\&times;P{\left({\mathrm{\&lambda;}}^k\right)}^{\mathrm{PSF}}+R+S}\right)}_i^{\mathrm{PSF}}$ (formula 4)

It should be noted that, can not need to define orthogonal projection conversion and back project function, directly PSF model integration is entered in image reconstruction algorithm formula 1.

Refer to the process flow diagram of the embodiment of Fig. 8 PET image rebuilding method of the present invention.

Step S11: the analog position data of obtaining calibration point according to Monte-carlo Simulation Method.

As shown in Figure 4, described calibration point is arranged in the 1/2 surface area F by the central point of PET gauging ring and the formed sector region of single detector module T of corresponding described central point, wherein, described PET gauging ring comprises a plurality of detector modules of arranging according to regular polygon.Described PET gauging ring can be the square being comprised of 4 detector modules, the regular hexagon being comprised of 6 detector modules, the octagon being comprised of 8 detector modules.It should be noted that, the quantity of the detector module that described PET gauging ring is included is not limited to the cited quantity of the present invention.

Step S12: the analog position data of described calibration point are rearranged, obtain the sinogram of described calibration point.

On the position at every LOR place, have a lot of events, the event of all identical LOR position is merged, and according to the X of LOR _rrearrange with φ, obtain sinogram.

Step S13: the sinogram of described calibration point is carried out to interpolation arithmetic, to proofread and correct described sinogram.

Original sinogram is at X _rinhomogeneous with φ interval, in order to meet the needs of subsequent reconstruction and modeling, original sinogram is changed into the sinogram of interpolation.Concrete grammar as shown in Figure 6, is projected as example with certain, and black real point represents the position of LOR in original sinogram, and the spacing between them is inhomogeneous, and hollow dots represents the result after interpolation, and they are evenly spaced.The interpolation point data of hollow dots can be obtained by linear interpolation by the black real point data of left and right.

Step S14: according to described sinogram, build image reconstruction model formation 1 or formula 2 based on maximum likelihood expectation maximization algorithm;

Step S15: set up the PSF model shown in formula 2, and formula 2 is integrated in formula 1 or formula 2, obtain formula 4;

Step S16: measured data is inputted to formula 4, obtain PET and rebuild image.

As Fig. 5 by Fig. 4 by the central point of PET gauging ring with corresponding as described in 1/2 surface area of the formed sector region of single detector module of central point amplify, the calibration point position of stain for calculating in figure, R is gauging ring radius.Calibration point is uniformly distributed in Euler space.By one of them eccentric calibration point, can obtain the sinogram on Fig. 5 the right.In general, near X _rthe distribution of=0 o'clock PSF has less half-peak breadth (Full Width at Half Maximum, FWHM), at X _rwhen bias is larger, PSF often has larger half-peak breadth.In view of being distributed in about peak position of PSF has asymmetry, therefore set up PSF model and proofread and correct.

Refer to the schematic diagram of Fig. 9 PET image re-construction system of the present invention.

Described PET image re-construction system 100 at least comprises:

Data of calibration point acquisition device 101, for obtaining the data of calibration point of PET image;

Image reconstruction model construction device 102, for according to the statistical nature of described data of calibration point, builds the image reconstruction model based on maximum likelihood expectation maximization algorithm;

Image reconstruction model correcting device 103, for utilizing point spread function to revise described image reconstruction model, obtains revised image reconstruction model;

PET rebuilds image calculation device 104, for measured data being inputted to described revised image reconstruction model, obtains PET and rebuilds image.

Preferably, described data of calibration point acquisition device 101 comprises:

System environments virtual module 1011, for the virtual point source that builds virtual pet system environments and be contained in described virtual pet system environments;

Data of calibration point analog module 1012, for scanning described virtual point source in described virtual pet system environments, obtains the position data of calibration point;

Sinogram generation module 1013, for generating the sinogram of described calibration point according to the position data of described calibration point.

Preferably, described calibration point is arranged in 1/2 surface area by the central point of PET gauging ring and the formed sector region of single detector module of corresponding described central point, wherein, described PET gauging ring comprises a plurality of detector modules of arranging according to regular polygon.

Preferably, also comprise sinogram correction module 1014, for the sinogram of described calibration point is carried out to interpolation arithmetic, to proofread and correct described sinogram.

Preferably, described image reconstruction model construction device 102 comprises:

Default image collection module 1021, for obtaining a default image;

Estimate sinogram computing module 1022, for calculating described default image along the projection of each φ direction, obtain estimating sinogram, wherein, φ is through the light of radioactive point source and the angle of horizontal direction in PET system;

Comparison module 1023, for relatively estimating the error between sinogram and actual measurement sinogram, obtains error sinogram;

Interative computation module 1024, for described error sinogram is carried out to back projection, to upgrade estimated image, the row iteration computing of going forward side by side.

Preferably, the formula of described interative computation is:

${\mathrm{\&lambda;}}_i^{k+1}=\frac{{\mathrm{\&lambda;}}_i^k}{{\mathrm{\&Sigma;}}_{m=1}^M{F}_{\mathrm{mi}}{A}_m}\&times;\underset{j=1}{\overset{M}{\mathrm{\&Sigma;}}}{F}_{\mathrm{ji}}\left(\frac{A_j{p}_j}{A_j{\mathrm{\&Sigma;}}_{l=1}^N{F}_{\mathrm{jl}}{\mathrm{\&lambda;}}_l^k+R_j+S_j}\right)$

Wherein, the value of i pixel of presentation video after the k time iteration, p _jthe collection value that represents j data point in sinogram, A, R, S represent respectively attenuation correction coefficient, random coincidence event number, scattering events number, F _jithe probability that the event that expression produces from i pixel collects in j data point of sinogram, M represents the number of sinogram spatial data points, the number of N presentation video pixel.

Preferably, described image reconstruction model correcting device 103 comprises:

PSF model building module 1031, for setting up PSF model, described PSF model is:

$\mathrm{PSF}=\exp (-\frac{{(\mathrm{xr}-x_0)}^2}{{2\mathrm{\&sigma;}}^2}),$ $\mathrm{\&sigma;}=\left\{\begin{array}{cc}{\mathrm{\&sigma;}}_{\mathrm{left}}& \mathrm{xr}<x_0\\ {\mathrm{\&sigma;}}_{\mathrm{right}}& \mathrm{xr}>=x_0\end{array}\right.;$ Wherein, X _rfor departing from centre distance, x ₀x corresponding to peak position estimating _rposition, σ _leftand σ _rightbe respectively x ₀the distribution variance on the left side and the right;

PSF correcting module 1032, for by described PSF model substitution image reconstruction model, obtains revised image reconstruction model.

The explanation of system embodiment please refer to embodiment of the method, and the present invention does not repeat them here.

In sum, a kind of PET image rebuilding method of the present invention and system, have following beneficial effect:

First, the present invention adopts maximum likelihood expectation maximization algorithm and the correction of PSF model is combined, and the process of image reconstruction is carried out in sinogram (Sinogram) space, and algorithm complex is by traditional O (N ⁶) be reduced to O (N ⁴), reduced reconstruction operation time.

Secondly, utilize the symmetry in sinogram space, make the calibration point of experiment only need to cover 1/2 surface area by the central point of PET gauging ring and the formed sector region of single detector module of corresponding described central point, thereby the quantity of calibration point is greatly reduced, reduced the demand to storage, computational resource.

Again, the method by Monte Carlo (Monte-Carlo) simulation obtains data of calibration point, without relying on positioning mechanical system, has avoided the error causing due to a Source size and mechanical positioning precision.

So the present invention has effectively overcome various shortcoming of the prior art and tool high industrial utilization.

Above-described embodiment is illustrative principle of the present invention and effect thereof only, but not for limiting the present invention.Any person skilled in the art scholar all can, under spirit of the present invention and category, modify or change above-described embodiment.Therefore, such as in affiliated technical field, have and conventionally know that the knowledgeable, not departing from all equivalence modifications that complete under disclosed spirit and technological thought or changing, must be contained by claim of the present invention.

Claims (15)

1. a PET image rebuilding method, is characterized in that, described PET image rebuilding method at least comprises:

Obtain the data of calibration point of PET image;

According to the statistical nature of described data of calibration point, build the image reconstruction model based on maximum likelihood expectation maximization algorithm;

Utilize point spread function to revise described image reconstruction model, obtain revised image reconstruction model;

Measured data is inputted to described revised image reconstruction model, obtain PET and rebuild image.

2. PET image rebuilding method according to claim 1, is characterized in that, the data of calibration point of obtaining PET image further comprises:

Build virtual pet system environments and be contained in the virtual point source in described virtual pet system environments;

In described virtual pet system environments, scan described virtual point source, obtain the position data of calibration point;

According to the position data of described calibration point, generate the sinogram of described calibration point.

3. PET image rebuilding method according to claim 2, it is characterized in that: described calibration point is arranged in 1/2 surface area by the central point of PET gauging ring and the formed sector region of single detector module of corresponding described central point, wherein, described PET gauging ring comprises a plurality of detector modules of arranging according to regular polygon.

4. PET image rebuilding method according to claim 2, is characterized in that, after the sinogram that generates described calibration point according to the position data of described calibration point, also comprises: the sinogram of described calibration point is carried out to interpolation arithmetic, to proofread and correct described sinogram.

5. PET image rebuilding method according to claim 1, is characterized in that: the image reconstruction model building based on maximum likelihood expectation maximization algorithm further comprises:

Obtain a default image;

Calculate described default image along the projection of each φ direction, obtain estimating sinogram, wherein, φ is through the light of radioactive point source and the angle of horizontal direction in PET system;

Relatively estimate the error between sinogram and actual measurement sinogram, obtain error sinogram;

Described error sinogram is carried out to back projection, to upgrade estimated image, the row iteration computing of going forward side by side.

6. PET image rebuilding method according to claim 5, is characterized in that: the formula of described interative computation is:

Wherein, the value of i pixel of presentation video after the k time iteration, p _jthe collection value that represents j data point in sinogram, A, R, S represent respectively attenuation correction coefficient, random coincidence event number, scattering events number, F _jithe probability that the event that expression produces from i pixel collects in j data point of sinogram, M represents the number of sinogram spatial data points, the number of N presentation video pixel.

7. PET image rebuilding method according to claim 1, is characterized in that, utilizes point spread function to revise described image reconstruction model, obtains revised image reconstruction model and further comprises:

Set up PSF model, described PSF model is:

wherein, X _rfor departing from centre distance, x ₀x corresponding to peak position estimating _rposition, σ _leftand σ _rightbe respectively x ₀the distribution variance on the left side and the right;

By described PSF model substitution image reconstruction model, obtain revised image reconstruction model.

8. PET image rebuilding method according to claim 7, is characterized in that, by described PSF model substitution image reconstruction model, obtains revised image reconstruction model and further comprises:

The orthogonal projection transforming function transformation function of definition image reconstruction model is P () _j=∑ _if _ji(), back project function is BP () _i=∑ _jf _ji(), utilizes described PSF model to align projective transformation function and back project function is revised, and obtains revised orthogonal projection transforming function transformation function back project function

By described revised orthogonal projection transforming function transformation function and back project function substitution image reconstruction model, obtain revised image reconstruction model.

9. a PET image re-construction system, is characterized in that, described PET image re-construction system at least comprises:

Data of calibration point acquisition device, for obtaining the data of calibration point of PET image;

Image reconstruction model construction device, for according to the statistical nature of described data of calibration point, builds the image reconstruction model based on maximum likelihood expectation maximization algorithm;

Image reconstruction model correcting device, for utilizing point spread function to revise described image reconstruction model, obtains revised image reconstruction model;

PET rebuilds image calculation device, for measured data being inputted to described revised image reconstruction model, obtains PET and rebuilds image.

10. PET image re-construction system according to claim 9, it is characterized in that, described data of calibration point acquisition device comprises: system environments virtual module, for the virtual point source that builds virtual pet system environments and be contained in described virtual pet system environments;

Data of calibration point analog module, for scanning described virtual point source in described virtual pet system environments, obtains the position data of calibration point;

Sinogram generation module, for generating the sinogram of described calibration point according to the position data of described calibration point.

11. PET image re-construction systems according to claim 10, it is characterized in that: described calibration point is arranged in 1/2 surface area by the central point of PET gauging ring and the formed sector region of single detector module of corresponding described central point, wherein, described PET gauging ring comprises a plurality of detector modules of arranging according to regular polygon.

12. PET image re-construction systems according to claim 10, is characterized in that: also comprise sinogram correction module, for the sinogram of described calibration point is carried out to interpolation arithmetic, to proofread and correct described sinogram.

13. PET image re-construction systems according to claim 9, is characterized in that, described image reconstruction model construction device comprises:

Default image collection module, for obtaining a default image;

Estimate sinogram computing module, for calculating described default image along the projection of each φ direction, obtain estimating sinogram, wherein, φ is through the light of radioactive point source and the angle of horizontal direction in PET system;

Comparison module, for relatively estimating the error between sinogram and actual measurement sinogram, obtains error sinogram;

Interative computation module, for described error sinogram is carried out to back projection, to upgrade estimated image, the row iteration computing of going forward side by side.

14. PET image re-construction systems according to claim 13, is characterized in that: the formula of described interative computation is:

Wherein, the value of i pixel of presentation video after the k time iteration, p _jthe collection value that represents j data point in sinogram, A, R, S represent respectively attenuation correction coefficient, random coincidence event number, scattering events number, F _jithe probability that the event that expression produces from i pixel collects in j data point of sinogram, M represents the number of sinogram spatial data points, the number of N presentation video pixel.

15. PET image re-construction systems according to claim 9, is characterized in that, described image reconstruction model correcting device comprises:

PSF model building module, for setting up PSF model, described PSF model is:

wherein, X _rfor departing from centre distance, x ₀x corresponding to peak position estimating _rposition, σ _leftand σ _rightbe respectively x ₀the distribution variance on the left side and the right;

PSF correcting module, for by described PSF model substitution image reconstruction model, obtains revised image reconstruction model.