CN104408756B

CN104408756B - A kind of PET image reconstruction method and device

Info

Publication number: CN104408756B
Application number: CN201410603993.1A
Authority: CN
Inventors: 赵大哲; 王军搏; 韩冬; 栗伟; 耿欢
Original assignee: Neusoft Corp
Current assignee: Neusoft Corp
Priority date: 2014-10-30
Filing date: 2014-10-30
Publication date: 2017-05-31
Anticipated expiration: 2034-10-30
Also published as: CN104408756A

Abstract

The invention discloses a kind of PET image reconstruction method, including：According to the table schema list-mode data rebuild for PET, determine respectively it is each in the list-mode data meet the corresponding spatial encoding data of event data, the spatial encoding data includes the first coding, the second coding, the 3rd coding, the 4th coding and the 5th coding；All spatial encoding datas are ranked up according to pre-set space order；Each spatial encoding data is processed successively according to ranking results, PET image reconstruction is completed with by iterative reconstruction algorithm.The invention also discloses a kind of PET image reconstruction device.

Description

A kind of PET image reconstruction method and device

Technical field

The present invention relates to technical field of image processing, more particularly to a kind of PET image reconstruction method and device.

Background technology

Currently, positron emission tomography (Positron Emission Computed Tomography, abbreviation PET) technology has been widely used in the important field of biomedical research such as tumour early detection, drug screening, and traditional Medical imaging modalities are compared to obvious advantage.PET technologies are shown by radioactivity of the injection with targeting in human body Track agent, and annihilation photon signal detection is carried out in vitro, PET can reconstruct the spatial distribution of tracer, so that indirect gain The physiological metabolism information of human body.And being based on different data mode and method for reconstructing often represents PET image reconstruction effects Go out obvious difference, the data memory format for being commonly used for PET reconstructions is divided into sinogram in units of data for projection (Sinogram) data and table schema (list-mode) data in units of meeting event.

Due to sinogram data storage is data for projection, and its data volume can increase as the quantity of pet detector increases Plus, when flight time (Time of Flight, the abbreviation TOF) information of combination, sinogram data amount can also improve decades of times, weight Building the time can also increase therewith.To solve this problem, conventional method is to use projection angle merging, the axle in fault plane Sinogram data is reduced to the mode of the reduction data samplings such as ring difference merging to the demand of memory space and iteration calculation is improved The reconstruction speed of method.However, these methods can cause reconstruction image tangential spatial resolution or axial space in fault plane The different degrees of decline of resolution ratio.And be directed to TOF-PET and rebuild and then will also meet event time difference and be divided into some larger Away from time interval (time-bin) down-sampled treatment is carried out to temporal information, limit TOF and rebuild to imaging signal to noise ratio Raising ability.

Compared to the data form of sinogram, the data of List-mode forms then can be good at solving the above problems. List-mode data forms, are that the time sequencing for meeting event is detected according to PET system, by the time of annihilation photon, energy The data mode that the information such as amount, position are stored successively, the size of its data volume only to meet event count number it is related, not Can increase with the increase of detector cells quantity.Meanwhile, list-mode data can naturally with TOF method for reconstructing knots Close, finer time-bin can be carried out and divided, so as to remain all information of initial data, make further to improve data Picking rate, reduction drug dose, improve PET image quality and are possibly realized.

However, the method for reconstructing based on list-mode data is also faced with, reconstruction speed is slow, reconstruction time is with meeting thing Part count increase and substantially increase these problems.The conventional iterative method for reconstructing of List-mode data, is thrown using sinogram Maximum likelihood expectation maximization (the Maximum Likelihood Expectation Maximization, abbreviation of shadow data MLEM) algorithm for reconstructing is extended in the single PET reconstructions for meeting event, but the reconstruction speed of this method is delayed very much Slowly, it is impossible to directly clinically apply.At present, the iteratively faster of list-mode data rebuilds use for reference based on sinogram projection more Order subset expectation maximization method (the Ordered Subsets Expectation Maximization, abbreviation of data OSEM), divided according to the time sequencing of collection by by list-mode data, so as to realize based on time subset OSEM iteratively fasters are rebuild.Because when TOF rebuilds, every projection line can only have small part to act on image space, and Need for whole piece projection line to be applied to image space unlike non-TOF rebuilds, projection angle merging, axial direction are used in TOF reconstructions When speed is rebuild in the down-sampled method reduction data space such as ring difference merging and raising, the space in respective direction can be reduced Resolution loss, therefore, be applied to this kind of method during the TOF of list-mode data rebuilds, and by same sky by prior art Between the event that meets of position merged, reconstruction time is not met the increased influence of event count.Additionally, existing The use data arrangement method similar to sinogram data reconstruction, will list-mode data drawn according to projecting direction Point, and rebuild using OSEM algorithms, it is suitable that this method makes list-mode data spatially realize to a certain extent Sequence is grouped, and reconstruction speed more faster than the OSEM methods of use time ordered subsets can be realized in theory.Except rebuilding calculation Optimization and improvement in method, prior art is also based on graphics processing unit (Graphic Processing Unit, abbreviation GPU hardware-accelerated method for reconstructing aspect) has been also carried out in-depth study, and preferable effect is served in terms of accelerated reconstruction.

The previously described fast reconstructing method based on algorithm optimization can play certain adding in image reconstruction Speed effect, but these methods still suffer from obvious deficiency in reconstruction speed.

The content of the invention

In view of this, the main purpose of the embodiment of the present invention is to provide a kind of PET image reconstruction method and device, with reality Now improve the purpose of image reconstruction speed.

To achieve the above object, a kind of PET image reconstruction method is the embodiment of the invention provides, including：

According to the table schema list-mode data rebuild for PET, determine respectively each in the list-mode data Meet the corresponding spatial encoding data of event data, the spatial encoding data includes that the first coding, the second coding, the 3rd compile Code, the 4th coding and the 5th coding；

All spatial encoding datas are ranked up according to pre-set space order；

Each spatial encoding data is processed successively according to ranking results, PET is completed with by iterative reconstruction algorithm Image reconstruction；

Wherein,

Described first is encoded to and meets the corresponding direction encoding of projecting direction of the event in fault plane, and the tomography is put down Face is the annular section of detector rings；

Described second is encoded to and described meets the corresponding regional code of view field of the event under the projecting direction, institute It is to detect the region met between two probe units of event to state view field；

Described 3rd is encoded to and described meets the corresponding ring difference coding of detector rings difference that event has, the detector Ring difference is the difference with 1 after number of rings, it is described after number of rings be the view field on fault plane axial direction after Detector rings quantity；

Described 4th is encoded to the corresponding time-bin codings in time-bin positions for meeting event described in generation, described Time-bin positions be along the projecting direction in view field according to list-mode data time difference information divide it is each One in individual time interval；

Described 5th is encoded to the corresponding axial direction of axial location of the event in the time-bin positions that meets compiles Code.

Preferably, it is used for the list-mode data that PET rebuilds in the basis, the list-mode data is determined respectively In it is each meet the corresponding spatial encoding data of event data before, also include：

If each detector rings includes M probe unit, projecting direction of the event in the fault plane will be met It is divided intoPlant or M kinds, and every kind of projecting direction is encoded to obtainIndividual or M direction encoding；

When projecting direction is divided intoWhen planting, by M-1 projected area under same projection direction in the fault plane Domain is encoded, to obtain M-1 regional code, when projecting direction is divided into M kinds, by identical throwing in the fault plane Under shadow directionIt is individual orIndividual view field is encoded, to obtainIt is individual orIndividual regional code；

If in the presence of P detector rings being sequentially placed side by side, on fault plane axial direction, by same projection direction Under Q kind detector rings differences encoded, to obtain Q ring difference coding, Q≤P；

Fault plane axial region is divided into the N number of time interval axially in parallel with fault plane, and to same projection N number of time interval under direction is encoded, and is encoded with obtaining N number of time-bin；

The Z kind axis projections region with same probe ring difference RD under to same projection direction carries out axial coding, To obtain Z axially coding, Z=P-RD；

Wherein, M,Q, P, N, Z are natural number.

Preferably, it is described that all spatial encoding datas are ranked up according to pre-set space order, including：

Same first data group will be divided into all spatial encoding datas that there should be equidirectional to encode, and by each institute The first data group is stated to be ranked up according to the first preset order；

By in first data group to should have same area encode all spatial encoding datas be divided into same second Data group, and each described second data group is ranked up according to the second preset order；

The same 3rd will be divided into all spatial encoding datas that there should be identical ring difference coding in second data group Data group, and each described 3rd data group is ranked up according to the 3rd preset order；

It is same by being divided into all spatial encoding datas that should have identical time-bin coding in the 3rd data group 4th data group, and each described 4th data group is ranked up according to the 4th preset order；

All spatial encoding datas in 4th data group are ranked up according to the 5th preset order.Preferably, It is described that each spatial encoding data is processed successively according to ranking results, including：

Obtain each spatial encoding data for meeting event successively according to ranking results；If the current space for meeting event is compiled Code data are different from the spatial encoding data that upper meets event, then meet with reference to flight time TOF information and currently event Spatial encoding data carry out before to and backwards projection calculate and currently meet event forward projection and backwards projection result obtaining；If The current spatial encoding data for meeting event is identical with the spatial encoding data that upper meets event, then before meeting event by upper one To projection or backwards projection result as currently meeting event forward projection or backwards projection result；

Or, obtaining each spatial encoding data for meeting event successively according to ranking results；For meeting for current acquisition The spatial encoding data of event, with reference to TOF information and currently meet event spatial encoding data carry out before to and backwards projection Calculating currently meets event forward projection and backwards projection result to obtain.

Preferably, methods described also includes：

Each described first data group is respectively stored into each the first contiguous memory space, and to each first it is continuous in Space is deposited to be numbered according to first preset order；

Each described second data group is respectively stored into each the second contiguous memory space, and to each second it is continuous in Space is deposited to be numbered according to second preset order；

Each described 3rd data group is respectively stored into each the 3rd contiguous memory space, and to each the 3rd it is continuous in Space is deposited to be numbered according to the 3rd preset order；

Each described 4th data group is respectively stored into each the 4th contiguous memory space, and to each the 4th it is continuous in Space is deposited to be numbered according to the 4th preset order；

Each spatial encoding data in 4th data group is respectively stored into each the 5th contiguous memory space, and Each the 5th contiguous memory space is numbered according to the 5th preset order；

And/or,

Each described first data group is respectively stored into each the first contiguous memory space according to direction encoding, and to relating to And all directions coding be numbered according to first preset order；

Each described second data group is respectively stored into each the second contiguous memory space according to regional code, and to relating to And regional coding be numbered according to second preset order；

Each described 3rd data group is respectively stored into each the 3rd contiguous memory space according to ring difference coding, and to relating to And each ring difference coding be numbered according to the 3rd preset order；

Each described 4th data group is respectively stored into each the 4th contiguous memory space according to time-bin codings, and Each time-bin codings to being related to are numbered according to the 4th preset order；

Each spatial encoding data in the 4th data group is respectively stored into each the 5th company according to axially coding Continuous memory headroom, and to be related to each axially coding be numbered according to the 5th preset order；

Wherein, the second contiguous memory space belongs to the first contiguous memory space, and the 3rd contiguous memory is empty Between belong to the second contiguous memory space, the 4th contiguous memory space belongs to the 3rd contiguous memory space, described 5th contiguous memory space belongs to the 4th contiguous memory space.

It is described to obtain each spatial encoding data for meeting event successively according to ranking results, including：

Numbered according to memory headroom and/or to the numbering of space encoding, obtained in memory headroom successively according to number order Each spatial encoding data for meeting event.

The embodiment of the present invention additionally provides a kind of PET image reconstruction device, including：

Coding determining unit, for according to the table schema list-mode data rebuild for PET, determining respectively described Each in list-mode data to meet the corresponding spatial encoding data of event data, the spatial encoding data includes that first compiles Code, the second coding, the 3rd coding, the 4th coding and the 5th coding；

Event ordering unit, for all spatial encoding datas to be ranked up according to pre-set space order；

Image reconstruction unit, for the ranking results according to the event ordering unit successively to each spatial encoding data Processed, PET image reconstruction is completed with by iterative reconstruction algorithm；

Wherein,

Preferably, described device also includes：

Direction encoding unit, according to the list-mode data rebuild for PET, divides in the coding determining unit Do not determine in the list-mode data it is each meet the corresponding spatial encoding data of event data before, if each detector rings Comprising M probe unit, then will meet projecting direction of the event in the fault plane and be divided intoPlant or M kinds, and to every Projecting direction is planted to be encoded to obtainIndividual or M direction encoding；Regional code unit, for being divided into when projecting directionWhen planting, M-1 view field under same projection direction in the fault plane is encoded, to obtain M-1 region Coding, when projecting direction is divided into M kinds, by under same projection direction in the fault planeIt is individual orIndividual throwing Shadow zone domain is encoded, to obtainIt is individual orIndividual regional code；

Ring difference coding unit, if in the presence of P detector rings being sequentially placed side by side, in fault plane axial direction On, the Q kind detector rings differences under same projection direction are encoded, to obtain Q ring difference coding, Q≤P；

Time encoding unit, for fault plane axial region to be divided into the N number of time axially in parallel with fault plane Interval, and to same projection direction under N number of time interval encode, encoded with obtaining N number of time-bin；

Axial coding unit, for the Z kind axis projections with same probe ring difference RD under to same projection direction Region carries out axial coding, to obtain Z axially coding, Z=P-RD；

Wherein, M,Q, P, N, Z are natural number.

Preferably, the event ordering unit, including：

First sequence subelement, for same the will to be divided into all spatial encoding datas that should have equidirectional to encode One data group, and each described first data group is ranked up according to the first preset order；

Second sequence subelement, for by first data group to should have same area encode all space encodings Data are divided into same second data group, and each described second data group is ranked up according to the second preset order；

3rd sequence subelement, for by all space encodings in second data group to that should have identical ring difference coding Data are divided into same 3rd data group, and each described 3rd data group is ranked up according to the 3rd preset order；

4th sequence subelement, for by all spaces in the 3rd data group to that should have identical time-bin coding Coded data is divided into same 4th data group, and each described 4th data group is ranked up according to the 4th preset order；

5th sequence subelement, for all spatial encoding datas in the 4th data group are default suitable according to the 5th Sequence is ranked up.

Preferably, described image reconstruction unit, including：

Event data obtains subelement, for obtaining each space encoding number for meeting event successively according to ranking results According to；

First iterative calculation subelement, if for currently meeting the sky that the spatial encoding data of event meets event with upper Between coded data it is different, then with reference to flight time TOF information and currently meet event spatial encoding data carry out before to and it is inverse Currently meet event forward projection and backwards projection result to obtain to projection calculating；

Iteration result assignment subelement, if for currently meeting the sky that the spatial encoding data of event meets event with upper Between coded data it is identical, then meet event forward projection or backwards projection result as currently meeting event forward projection using upper one Or backwards projection result；

Or, described image reconstruction unit, including：

Secondary iteration computation subunit, for for the current spatial encoding data for meeting event for obtaining, with reference to TOF letters Breath and currently meet event spatial encoding data carry out before to and backwards projection calculate to obtain current meeting before event to throwing Shadow and backwards projection result.

Preferably, described device also includes：

First space numbered cell, it is empty for each described first data group to be respectively stored into each first contiguous memory Between, and each the first contiguous memory space is numbered according to first preset order；

Second space numbered cell, it is empty for each described second data group to be respectively stored into each second contiguous memory Between, and each the second contiguous memory space is numbered according to second preset order；

3rd space numbered cell, it is empty for each described 3rd data group to be respectively stored into each the 3rd contiguous memory Between, and each the 3rd contiguous memory space is numbered according to the 3rd preset order；

4th space numbered cell, it is empty for each described 4th data group to be respectively stored into each the 4th contiguous memory Between, and each the 4th contiguous memory space is numbered according to the 4th preset order；

5th space numbered cell, for each spatial encoding data in the 4th data group to be respectively stored into respectively Individual 5th contiguous memory space, and each the 5th contiguous memory space is numbered according to the 5th preset order；

And/or,

First coding numbered cell, for according to direction encoding by each described first data group be respectively stored into each the One contiguous memory space, and to be related to all directions coding be numbered according to first preset order；

Second coding numbered cell, for according to regional code by each described second data group be respectively stored into each the Two contiguous memory spaces, and to be related to regional coding be numbered according to second preset order；

3rd coding numbered cell, for according to ring difference coding by each described 3rd data group be respectively stored into each the Three contiguous memory spaces, and to be related to each ring difference coding be numbered according to the 3rd preset order；

4th coding numbered cell, for being respectively stored into respectively each described 4th data group according to time-bin codings Individual 4th contiguous memory space, and to be related to each time-bin coding be numbered according to the 4th preset order；

5th coding numbered cell, for being encoded each spatial encoding data in the 4th data group according to axial direction Be respectively stored into each the 5th contiguous memory space, and to be related to each axially coding carried out according to the 5th preset order Numbering；

Wherein, the second contiguous memory space belongs to the first contiguous memory space, and the 3rd contiguous memory is empty Between belong to the second contiguous memory space, the 4th contiguous memory space belongs to the 3rd contiguous memory space, described 5th contiguous memory space belongs to the 4th contiguous memory space；

The event data obtains subelement, specifically for being numbered and/or to the numbering of space encoding according to memory headroom, Obtain each spatial encoding data for meeting event in memory headroom successively according to number order.

PET image reconstruction method provided in an embodiment of the present invention and device, by determining each symbol in list-mode data The corresponding spatial encoding data of event data is closed, and all spatial encoding datas for meeting event is ranked up, it is ensured that The complete ordering property in space of list-mode data, using the orderly list-mode spatial encoding datas of complete space, can be direct Space orientation is carried out to meeting event data, so as to need to only calculate non-zero, computing resource is not only saved, also added Fast reconstruction speed.

Additionally, this programme can judge that currently meet event space position meets with upper one according to the spatial encoding data Whether event space position is identical, directly is assigned to currently meet event by a upper result of calculation if identical, so that significantly Reduce and compute repeatedly the amount of calculation that same position meets event, it is to avoid identical data is computed repeatedly so that do not make During with the approximation method such as any data are down-sampled, the fast and accurately image reconstruction based on list-mode data is can be achieved with, Even if meeting event number to be multiplied, a small amount of reconstruction time only also can be increased.

Brief description of the drawings

In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are the present invention Some embodiments, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis These accompanying drawings obtain other accompanying drawings.

Fig. 1 is embodiment of the present invention List-mode data space coding schematic diagrams；

Fig. 2 is embodiment of the present invention projecting direction and the schematic diagram of view field first；

Fig. 3 is embodiment of the present invention projecting direction and the schematic diagram of view field second；

Fig. 4 is the schematic flow sheet of embodiment of the present invention PET image reconstruction method；

Fig. 5 is that embodiment of the present invention data group divides dendrogram；

Fig. 6 realizes flow chart for embodiment of the present invention iterative reconstruction algorithm；

Fig. 7 is the structural representation of embodiment of the present invention PET image reconstruction device.

Specific embodiment

To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is A part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.

On the premise of ensureing that PET image quality is unaffected, the embodiment of the present invention proposes a kind of based on complete space The PET image reconstruction method of orderly list-mode data, realizes quick three-dimensional iterative reconstruction, and the method is rebuild from raising The non-sequential data of speed is accessed and non-duplicate two principal elements of calculating are starting point, is carried out by list-mode data Order space is encoded to realize non-sequential data access and non-duplicate calculating and combine for example iterative reconstruction approach such as OSEM, Do not use any data lack sampling treatment to ensure while computational accuracy, realize the rapid image based on list-mode data Rebuild.

The embodiment of the present invention is specifically introduced below.

In order to obtain the list-mode data of spacial ordering, it is necessary to the spatial information of list-mode data and time are believed Breath is completely converted into spatial encoding data, i.e., meet spatial information and the time difference of event in list-mode data by calculating Information acquisition can represent the group index coding for meeting event data complete information.This process refers to determine respectively each Meet the space encoding of event, i.e., meet event for a certain, according to the detector that annihilation photon is received in list-mode data Positional information and detector receive two time difference informations of annihilation photon, calculate this and meet projection side belonging to event To, meet view field position of the event under the direction, meet that the affiliated ring of event is poor, meet the affiliated time interval of event (time-bin) position and meet axial location of the event in affiliated time-bin positions distinguish it is corresponding five index compile Code parameter.

In order to obtain each space encoding information (i.e. above-mentioned five indexes coding parameter) for meeting event, before this, Need to predefine various coding situations, List-mode data space coding schematic diagrams shown in Figure 1, below as following Order defines five classes index coding successively：

1st, direction encoding

In embodiments of the present invention, if each detector rings includes M probe unit, event will be met in fault plane Projecting direction in (detector rings interior zone) is divided intoPlant or M kinds, and every kind of projecting direction is encoded to obtainOr M kind direction encodings；Wherein,M is natural number.

Specifically, referring to Fig. 1 (a), outside annular region is used for representing pet detector ring, two in detector rings One view field of Regional Representative (for example shadow region) between ray.If it is single that each pet detector ring contains M detection Unit, then use(wherein,) projection angle is represented, to meet paired annihilation photon in event be reverse fortune due to each Dynamic, so being projection angle range with 0 to 180 degree, optional position can be taken here for 0 degree of projection angle.Due to each throwing Shadow angle just corresponds to a projecting direction, can be divided into projecting directionPlant or M kinds, M is usually even number, when projection side It is to sumIt is individual, it is settable according to projecting directionThisIndividual direction encoding, when projecting direction sum It is M, φ=0 is can be set according to projecting direction, 1,2..., M-1 this M direction encoding.It should be noted that Fig. 1 (a) is Schematically, all view fields under all projecting directions and certain projecting direction are not put on display.

2nd, regional code

In embodiments of the present invention, when projecting direction is divided intoWhen planting, by same projection side in the fault plane M-1 downward view field is encoded, to obtain M-1 regional code, when projecting direction is divided into M kinds, by institute State in fault plane under same projection directionIt is individual orIndividual view field is encoded, to obtainIt is individual or Individual regional code；Wherein, M is natural number.

Specifically, referring to Fig. 1 (b), in fault plane, there is the view field under one group of same projection direction, use Bin (Bin ∈ N) represents the regional code of a certain view field under a certain projecting direction, when projecting direction is divided intoWhen planting, Can obtain the Bin=0 under same projecting direction, 1,2 ..., M-2 this M-1 regional code, when projecting direction is divided into M kinds, Can obtain under same projecting directionThisIt is individual orThisCompile in individual region Code,.It should be noted that Fig. 1 (b) is schematical, all view fields under certain projecting direction are not put on display.

In order to more easily illustrate the dividing mode of above-mentioned direction encoding and regional code, it is exemplified below：

Projecting direction and projected area shown in projecting direction shown in Figure 2 and the schematic diagram of view field first and Fig. 3 The schematic diagram of domain second, it is assumed here that the probe unit number M=8 of detector rings, each probe unit according to 0,1 ... 7 it is suitable Sequence is numbered, and the direction of arrow represents projecting direction, the region generation between region and dotted line and probe unit between dotted line Table view field.

In Fig. 2, such as probe unit 0, it is believed that probe unit 0 to probe unit 2 and probe unit 0 are single to detection Projecting direction between unit 3 is identical, and probe unit 0 to the projecting direction between probe unit 3 is taken herein, in this manner, Make the region between each two probe unit under every kind of projecting direction as a view field, divided in this projecting direction Under mode, co-exist inProjecting direction is planted, there is M-1 kinds view field, such as in Fig. 2 first under every kind of projecting direction Individual figure, M-1 view field under the projecting direction is respectively 12 regions, 03 region, 74 regions, 65 regions, 02 region, 73rd area Domain and 64 regions (02 region, 73 regions and 64 regions refer to first figure of Fig. 3).

In Fig. 3, such as probe unit 0, it is believed that probe unit 0 to probe unit 2 and probe unit 0 are single to detection Projecting direction between unit 3 is two kinds of projecting directions, in this manner, each two probe unit is made under some projecting directions Between region as a view field, the region between under some projecting directions there are three probe units is used as one View field, under this projecting direction dividing mode, for two kinds of adjacent projecting directions, its corresponding view field of difference Number is respectivelyWithKind.

3rd, ring difference coding

In embodiments of the present invention, if in the presence of P detector rings being sequentially placed side by side, in fault plane axial direction On, the Q kind detector rings differences under same projection direction are encoded, to obtain Q ring difference coding；Wherein, Q≤P, Q, P are Natural number.

Specifically, the detector rings difference RD is the difference with 1 after number of rings, wherein, described is in tomography after number of rings View field on plane axial direction after detector rings quantity, specifically, the detector rings difference can be visited by calculating The difference of detector rings coding where two probe units for meeting event is measured to obtain.For convenience of understanding, referring to Fig. 1 (c), For comprising the P pet detector of detector rings, (P=8, corresponds to detector rings 0,1 ... 7) respectively in figure, each ring difference coding Be respectively defined as RingDiff=0,1,2 ..., P-1 (RingDiff ∈ N), correspond respectively to detector rings difference RD for 0,1, The situation of 2 ... P-1.If for example, certain view field for meeting event sees positioned at 00th area of detector rings 0 in the axial direction Domain (solid line bar-shaped zone), then the view field after detector rings quantity be 1, therefore its detector rings difference be 0 so that should Meet the corresponding ring difference coding RingDiff=0 of event, likewise, view field is located at 11 regions, 22nd area of detector rings Domain ... 77 region meets event, and its corresponding detector rings difference is also 0, and then corresponding ring difference coding RingDiff is also 0；Again for example, if certain view field for meeting event sees positioned at 07 region (dashed region) of detector rings in the axial direction, Then the view field after detector rings quantity be 8, therefore its detector rings difference be 8-1=7 so that this meet event correspondence Ring difference coding RingDiff=7.

It should be noted that in actual use, Q kinds can be chosen from this P kinds detector rings difference, and it is compiled Code, you can encoded with the detector rings difference not to all kinds, the detector rings difference of a selected part species is encoded .

4th, time-bin codings

In embodiments of the present invention, fault plane axial region is divided into the N number of time axially in parallel with fault plane Interval, and to same projection direction under N number of time interval encode, encoded with obtaining N number of time-bin；Wherein, N is certainly So count.

Specifically, event is met for each, PET system can record two annihilation photons and reach two spies in opposite direction The time difference of device unit is surveyed, referring to Fig. 1 (d), the minimum interval t of the time difference is recorded according to PET system, can broken Layer plane axial region (region in figure between detector rings) be divided into several with t be interlude interval (time- Bin), so just can obtain in a certain projecting direction view field with t be interval N number of time interval (time-bin).If The time-bin quantity N=n+1 of division, if then representing that the time-bin for meeting event is encoded with Timebin (Timebin ∈ N), Then Timebin=0,1,2 ..., n.The time difference of detector is reached by using the annihilation photon recorded in list-mode data The time-bin positions belonging to when this annihilation event (meeting event) occurs can just be calculated.

It should be noted that above-mentioned N number of time interval (time-bin) divides according to minimum interval t, and The interval bigger than minimum interval t can essentially be selected carries out the division of time-bin, so plays certain acceleration The effect of calculating, but reconstructed image quality might have very slight decline.

5th, axially encode

In embodiments of the present invention, to same projection direction under the Z kind axis projections with same probe ring difference RD Region carries out axial coding, to obtain Z axially coding, Z=P-RD；Wherein, Z is natural number.

Specifically, after Timebin space encodings are completed, it is desirable to be determined for compliance with the unique positions of event, in addition it is also necessary to count Calculate and meet event location in one group of axially parallel view field.Referring to Fig. 1 (d), for comprising P spy Survey the pet detector of device ring, when detector rings difference RD for 0 when, there is view field that Z=P kinds are parallel to each other (i.e. in Fig. 1 (c) 8 solid line bar-shaped zones), when detector rings difference RD be 1 when, there is the view field ... ... that Z=P-1 kinds are parallel to each other, when When detector rings difference RD is P-1, there is Z=1 kinds view field.It is P=8, the situation of detector rings difference RD=2 in Fig. 1 (d), There is the view field that Z=8-2=6 kinds are parallel to each other in it.Same probe ring difference RD is represented with Z_Paral (Z_Paral ∈ N) Under the view field that is parallel to each other of Z=P-RD kinds carry out axial space coding, so for any one group of parallel projection region Axial coding range be Z_Paral=0,1,2..., P-RD.

So, just the spatial information of list-mode data and time can be believed by using this five spatial index codings Breath completely shows.

Be the schematic flow sheet of PET image reconstruction method provided in an embodiment of the present invention referring to Fig. 4, the method include with Lower step：

Step 401：According to the table schema list-mode data rebuild for PET, the list-mode numbers are determined respectively It is each in meet the corresponding spatial encoding data of event data, the spatial encoding data is encoded including first, second encodes, 3rd coding, the 4th coding and the 5th coding.

Wherein,

Before step 401, it has been explained that five definition modes of index coding parameter, on this basis, according to PET Each of system record meets the room and time information of event, can calculate respectively each meet event it is corresponding comprising this five Plant the space encoding information of index coding parameter.

In step 401, specifically determine in the following manner it is each meet the corresponding space encoding of event data, below tie Close Fig. 1 and related example is specifically described：

1st, according to the spatial information for meeting event in the list-mode data, it is determined that the event that meets is described disconnected Actual projecting direction in layer plane, and meet the corresponding direction volume of event data according to the actual projecting direction determines Code.

Assuming that PET system includes 8 detector rings, and each detector rings includes 32 probe units.

For meeting event A, the actual projecting direction for meeting event A can be calculated according to list-mode data, with reference to Fig. 1 (a), it is assumed that its actual projecting direction is the direction that solid arrow is pointed to, according to the definition mode of above-mentioned direction encoding, can be true The corresponding direction encoding φ of the fixed actual projecting direction, assume that φ=6 here.So, by calculating list-mode data In meet projecting direction of the event in fault plane, be numbered to meeting event data according to projecting direction, just can make Projecting direction identical meets event data in fault plane, and there is identical direction to number.

2nd, according to the spatial information for meeting event in the list-mode data, it is determined that the event that meets is in the reality Actual view field under the projecting direction of border, and meet the corresponding area of event data according to the actual view field determines Domain encodes.

For meeting event A, the actual view field for meeting event A can be calculated according to list-mode data, it is assumed that Actual view field is the shaded bar frame on the right in Fig. 1 (b), according to the definition mode that above-mentioned zone is encoded, it may be determined that the reality The corresponding regional code Bin of view field, it is assumed here that Bin=12.So, to meeting event with equidirectional numbering φ Data, can respectively determine its regional code Bin in fault plane.

3rd, according to the spatial information for meeting event in the list-mode data, it is determined that the event that meets is in the reality Actual detector ring under the projecting direction of border is poor, and meets event data correspondence according to actual detector ring difference determines Ring difference coding.

For meeting event A, poor, the knot that can calculate the actual detector ring that this meets event A according to list-mode data Fig. 1 (c) is closed, if detector number of rings P=8, meeting the detector rings difference RD=2 of event A, according to the definition side that above-mentioned ring difference is encoded Formula, it may be determined that meet the ring difference coding RingDiff=2 of event A.So, thing can be met to what is numbered with equidirectional respectively Number of packages evidence, its ring difference numbering in PET axial directions is determined according to spatial order.

4th, different detection lists are arrived separately at using two annihilation photons for meeting event described in the list-mode data The time difference of unit, it is determined that the real time met belonging to when event occurs is interval, and it is interval really according to the real time The fixed corresponding time-bin of event data that meets is encoded.

With reference to Fig. 1 (d), it is assumed that meeting in event A the time difference that two annihilation photons arrive separately at different probe units is 0, then it is middle time interval to meet the real time belonging to when event occurs interval, according to determining that above-mentioned time-bin is encoded Right way of conduct formula, it is assumed that time-bin coding ranges are

Timebin=0,1,2..., n=0,1,2 ... 6, then the interval corresponding time-bin of interlude be encoded to

Timebin=3.So, just will in list-mode data in the fault plane same projection direction and with phase Time-bin coding informations are converted to the time difference information for meeting event data of detector rings difference.

5th, according to the spatial information for meeting event in the list-mode data, it is determined that the event that meets is in the reality Actual axial view field in the time interval of border, and meet event data according to the actual axial direction view field determines Corresponding axially coding.

With reference to Fig. 1 (d), for meeting event A, can further in ring difference coding RingDiff according to list-mode data The actual axial view field of event A is determined for compliance with when=2 in the corresponding 6 kinds axis projection regions being parallel to each other, that is, is determined Particular location in meeting event A in the time interval corresponding to Timebin=3, according to the definition side of above-mentioned axially coding Formula, it is assumed that it is determined that actual axial view field be the 2nd article (dash area) in 6 kinds of view fields, then meet the axle of event A To coding Z_Paral=1.

It can be seen that, utilization orientation coding φ=6, regional code Bin=12, ring difference coding RingDiff=2, time-bin are compiled Code is Timebin=3, axially this five spatial indexs coding of coding Z_Paral=1 just can meet event by meet event A The spatial information and temporal information of data completely show.

Step 402：All spatial encoding datas are ranked up according to pre-set space order.

After step 401 is completed, and the spatial index of list-mode data is encoded, in addition it is also necessary to by list-mode spaces Coded data carries out order arrangement according to spatial distribution.Firstly the need of the event that is determined for compliance with index coding φ, Bin, Putting in order between RingDiff, Timebin and Z_Paral, it be not arbitrary that this puts in order, because different arrangements Mode may cause the temporal greatest differences of PET image reconstruction.By analysis and the checking of previous work, it is believed that be more beneficial for The space encoding indexed sequential for improving reconstruction speed is φ, Bin, RingDiff, Timebin and Z_Paral.

After the order of five entry indexs coding is confirmed, first, by all event space coded datas that meet according to direction φ points of coding is extremelyIn individual group of (or M), and by thisIndividual group of (or M) is ranked up according to certain order, wherein every group of space is compiled Code data all have identical projecting direction, i.e., in each group data to meet event projecting direction in fault plane homogeneous Together, in every group of data meet event all by the view field position under certain direction, meet that the affiliated ring of event is poor, meet event The axial location residing in time-bin with event is met corresponding each in affiliated time-bin positions encodes to represent, this Sample, only needs to be assured that wherein a certain meeting according to other the four entry indexs coding in addition to direction encoding in every group of data The relevant information of event.Then, respectively by each direction be grouped in spatial encoding data according to view field regional code Bin point to M-1 (Or) in individual group, and by this M-1 (Or) individual group be ranked up according to certain order.Again Then, the spatial encoding data during each view field is grouped respectively divides into P group according to ring difference coding RingDiff, and This P group is ranked up according to certain order.Subsequently, respectively by the spatial encoding data in each ring differential set according to In Timebin points to N number of group of time-bin codings, and this N number of group is ranked up according to certain order.Finally, respectively will be every Each spatial encoding data that Z_Paral is axially encoded with certain in individual time-bin packets is arranged according to certain order Sequence.So, finally realize from the list-mode data of time sequencing arrangement to the conversion of complete space ordered data collection.

In order to more easily understand the sortord of each spatial encoding data, drawn with reference to the data group shown in Fig. 5 Dendrogram is divided to illustrate：

1st, same first data group will be divided into all spatial encoding datas that should have equidirectional to encode, and by each First data group is ranked up according to the first preset order.

If as it was previously stated, direction encoding includesKind, then all spatial encoding datas (be able to may not at most be wrapped here Containing certain projecting direction) it is divided intoIndividual first data group, respectively the first data group 1, the first data group 2 ..., the first number According to groupAnd every one first data group includes all spatial encoding datas (such as Fig. 5) with same projection direction.By this According to the first preset order, for example direction encoding (from big to small or from small to large) is ranked up a little first data groups, it is assumed here that It is from left to right final ranking results.

In addition, each first data group can be respectively stored into each first contiguous memory sky by the embodiment of the present invention Between, and each the first contiguous memory space is numbered according to first preset order；And/or, will according to direction encoding Each described first data group is respectively stored into each the first contiguous memory space, and all directions to being related to are encoded according to institute The first preset order is stated to be numbered.Wherein, between each the first contiguous memory space can be mutually continuous memory headroom It can also be mutual discontinuous memory headroom.

It is understood that there is projecting direction identical spatial encoding data identical direction to compile in fault plane Number, there are spatial encoding datas that equidirectional is encoded to be divided into corresponding first data group for these, by by different the The storage of one data group is carried out according still further to projecting direction to the diverse location in Computer Storage space to the memory headroom of data storage Serial number so that the consistent spatial encoding data of projecting direction is in one piece of continuous Computer Storage space in fault plane In, so corresponding memory headroom can be found by a certain specific projection direction in fault plane and numbered, and then find the spy Determine all spatial encoding datas under projecting direction；It is of course also possible to according to projecting direction all directions being related to are encoded into Row serial number, can so find the numbering of corresponding direction encoding by a certain specific projection direction in fault plane, enter And find all spatial encoding datas under the specific projection direction.

2nd, same the will be divided into all spatial encoding datas that there should be same area to encode in first data group Two data groups, and each described second data group is ranked up according to the second preset order.

If as it was previously stated, the regional code of the view field under same projecting direction includesKind, then for the first data Group 1, can divide all spatial encoding datas in the first data group 1 at most (may not include certain view field here) ForIndividual second data group, respectively the second data group 11, the second data group 12 ..., the second data group 1And every 1 Two data groups include the spatial encoding data (such as Fig. 5) with same projection region under same projection direction.By these According to the second preset order, for example regional code (from big to small or from small to large) is ranked up two data groups.Similarly, other First data group is also grouped and sorts in the manner described above, will not be repeated here.

In addition, each second data group can be respectively stored into each second contiguous memory sky by the embodiment of the present invention Between, and each the second contiguous memory space is numbered according to second preset order；And/or, will according to regional code Each described second data group is respectively stored into each the second contiguous memory space, and regional to being related to is encoded according to institute The second preset order is stated to be numbered；Wherein, the second contiguous memory space belongs to the first contiguous memory space.

It is understood that all spatial encoding data storages of zone number identical under same projecting direction are arrived calculating The same area of machine memory space, carries out serial number so that tomography according still further to regional code to the memory headroom of data storage Plane inner region encodes consistent spatial encoding data and is in one piece of continuous Computer Storage space, so can be by disconnected Corresponding memory headroom numbering is found in a certain specific projection region in layer plane, and then finds all under the specific projection region Spatial encoding data；It is of course also possible to carry out order volume to the regional coding being related to according to view field in fault plane Number, so the numbering of corresponding regional code can be found by a certain specific projection region, and then find the specific projection area All spatial encoding datas under domain.

3rd, in second data group all spatial encoding datas that there should be identical ring difference coding will be divided into same the Three data groups, and each described 3rd data group is ranked up according to the 3rd preset order.

As it was previously stated, if the ring difference coding under same projecting direction includes P kinds, for the second data group 11, can be by All spatial encoding datas at most (here may be poor not comprising certain ring) in second data group 11 are divided into P the 3rd data Group, respectively the 3rd data group 111, the 3rd data group 112 ..., the 3rd data group 11P, and every one the 3rd data group includes The spatial encoding data (such as Fig. 5) with same probe ring difference under same projection direction.By these the 3rd data groups according to For example ring difference coding is ranked up 3rd preset order (from big to small or from small to large).Similarly, other second data groups It is grouped in the manner described above and sorts, will not be repeated here.

In addition, each the 3rd data group can be respectively stored into each the 3rd contiguous memory sky by the embodiment of the present invention Between, and each the 3rd contiguous memory space is numbered according to the 3rd preset order；And/or, will according to ring difference coding Each described 3rd data group is respectively stored into each the 3rd contiguous memory space, and to each ring difference coding for being related to according to institute The 3rd preset order is stated to be numbered；Wherein, the 3rd contiguous memory space belongs to the second contiguous memory space.

It is understood that all spatial encoding data storages of ring difference numbering identical under same projecting direction are arrived calculating The same area of machine memory space, carries out serial number so that tomography according still further to ring difference coding to the memory headroom of data storage The consistent spatial encoding data of plane inner ring difference coding is in one piece of continuous Computer Storage space, so can be by certain One specific detector ring difference finds corresponding memory headroom numbering, and then all spaces found under specific detector ring difference are compiled Code data；It is of course also possible to carry out serial number to each ring difference coding being related to according to detector rings difference, can so pass through A certain specific detector ring difference finds the numbering of corresponding ring difference coding, and then finds having time under specific detector ring difference Between coded data.

4th, all spatial encoding datas that there should be identical time-bin coding will be divided into together in the 3rd data group One the 4th data group, and each described 4th data group is ranked up according to the 4th preset order.

If as it was previously stated, under same projecting direction time-bin coding include N kinds, for the 3rd data group 111, can At most (certain may not included here and meet the first of event with by all spatial encoding datas in the 3rd data group 111 Beginning time-bin position) be divided into N number of 4th data group, respectively the 4th data group 1111, the 4th data group 1112 ..., Four data group 111N, and every one the 4th data group include under same projection direction with identical initial time-bin positions Spatial encoding data (such as Fig. 5).By these the 4th data groups, according to the 4th preset order, for example time-bin is encoded (from big to small Or from small to large) be ranked up.Similarly, other the 3rd data groups are also grouped and sort in the manner described above, no longer go to live in the household of one's in-laws on getting married herein State.

In addition, each the 4th data group can be respectively stored into each the 4th contiguous memory sky by the embodiment of the present invention Between, and each the 4th contiguous memory space is numbered according to the 4th preset order；And/or, compiled according to time-bin Each described 4th data group is respectively stored into each the 4th contiguous memory space, and each time-bin volumes to being related to by code Code is numbered according to the 4th preset order；Wherein, the 4th contiguous memory space belongs to the 3rd contiguous memory Space.

It is understood that the time-bin codings all spatial encoding data storages of identical under same projecting direction are arrived The same area in Computer Storage space, serial number is carried out according still further to time-bin codings to the memory headroom of data storage, So that time-bin encodes consistent spatial encoding data and is in one piece of continuous Computer Storage space, can so pass through Corresponding memory headroom numbering is found in the specific initial time-bin positions for meeting event, and then it is specific initial to find this All spatial encoding datas under time-bin positions；It is of course also possible to according to the initial time-bin positions to being related to Each time-bin codings carry out serial number, can so be found by a certain specific initial time-bin positions corresponding The numbering of time-bin codings, and then find all spatial encoding datas under the specific initial time-bin positions.

5th, all spatial encoding datas in the 4th data group are ranked up according to the 5th preset order.

Every 1 the 4th data group includes one or more spaces under identical ring difference coding and identical time-bin codings Coded data, the corresponding axially coding of these spatial encoding datas may it is all different, be likely to it is partly or entirely identical.If The corresponding axially coding of these spatial encoding datas is all different, then can be default suitable according to the 5th by these spatial encoding datas For example axially coding is ranked up sequence (from big to small or from small to large)；If the corresponding axially coding of these spatial encoding datas Part is identical and part is different, then these can be had into the spatial encoding data of axially different coding according to the 5th preset order For example axially encode the spatial encoding data that is ranked up (from big to small or from small to large) and will be encoded with same axial by Be ranked up according to data obtaining time (or other modes), referring to Fig. 5, such as, and spatial encoding data 11111 to 1111Z be by Arranged from big to small according to axially coding, wherein, axial direction coding of the data 11111 with 11112 is identical, because PET system is first obtained Get and get after data 11111 data 11112, thus data 1111 sort formerly, data 11112 sort rear；If this The corresponding axially coding of a little spatial encoding datas is all identical, then these can be had the space encoding number of same axial coding It is ranked up according to according to data obtaining time (or other modes).

In addition, each spatial encoding data in the 4th data group can be respectively stored into each by the embodiment of the present invention 5th contiguous memory space, and each the 5th contiguous memory space is numbered according to the 5th preset order；And/or, Each spatial encoding data in the 4th data group is respectively stored into each the 5th contiguous memory sky according to axially coding Between, and to be related to each axially coding be numbered according to the 5th preset order；Wherein, the 5th contiguous memory is empty Between belong to the 4th contiguous memory space, the 5th contiguous memory space is made up of one or more internal storage locations.

Based on above-mentioned sequence, final data sorting result is 11111,11112 ... 1111Z ..., i.e. last in Fig. 5 One row ranking results from left to right.

Step 403：Each spatial encoding data for meeting event is processed successively according to ranking results, with by repeatedly PET image reconstruction is completed for algorithm for reconstructing.

For the iterative reconstruction algorithm that step 403 is mentioned, implement as shown in fig. 6, the figure shows iteration such as OSEM Algorithm for reconstructing realizes basic procedure.Forward projection calculating is carried out to the initial estimation image for setting first, the reason obtained with it It is compared with actual measurement projection value by projection value, so as to obtain projection correction's factor, projection correction's factor is by back projection meter Image correction factor is obtained after calculation, after being acted on initial estimation image, an iteration calculating is just realized, be have passed through many Secondary iteration, estimates that image is gradually restrained, until it is corrected for meeting necessarily required reconstruction image.

Due to owning, it is necessary to be traveled through according to certain spatial order in the iterative approximation of existing sinogram data for projection Possible projection come before completing to and backwards projection calculate, and be mostly the neutral element in data space during this Calculated, wasted computing resource；And existing conventional list-mode data re-establishing methods can be only to nonzero element meter Calculate, but the unordered feature of its data space result in and rebuild the slow of speed.And complete space is orderly in using the embodiment of the present invention Data set carries out image reconstruction, then eliminate the defect of these two aspects.

In embodiments of the present invention, the one kind that specifically may be selected in following two ways realizes step 403：

Mode one：Obtain each spatial encoding data for meeting event successively according to ranking results；Judgement currently meets thing Whether the spatial encoding data that the spatial encoding data of part meets event from upper is different；If the current space encoding for meeting event Data are different from the spatial encoding data that upper meets event, then meet with reference to flight time TOF information and currently the sky of event Between coded data carry out before to and backwards projection calculate and currently meet event forward projection and backwards projection result obtaining；If working as Before meet event spatial encoding data it is identical with the spatial encoding data that upper meets event, then by upper one meet before event to Projection or backwards projection result are used as currently meeting event forward projection or backwards projection result.

Mode two：Obtain each spatial encoding data for meeting event successively according to ranking results；For current acquisition Meet the spatial encoding data of event, with reference to TOF information and currently meet event spatial encoding data carry out before to and it is reverse Projection calculating currently meets event forward projection and backwards projection result to obtain.

It is understood that for mode one, when the quantity of spatial encoding data is low data bulk and intermediate data amount During situation, due to meet event space coding repeat ratio very little, so judge continuous two meet event code whether phase Same and projection calculates this part operation, and its castering action to calculating speed is no or little, only in spatial encoding data Data volume it is larger when, due to space encoding repeat ratio gradually increase, so acceleration could be embodied gradually.It is right In mode two, whether not judging continuous two, to meet event code identical, and to meet event space coded data equal for each Projection calculating treatment need to be one by one carried out, when the quantity of spatial encoding data is the situation of low data bulk and intermediate data amount, Can in selection way two, its treatment effect gap compared with mode one is little.

In above two mode, in order to obtain each spatial encoding data successively according to ranking results, specifically can basis Memory headroom is numbered and/or to the numbering of space encoding, and obtained successively according to number order in memory headroom each meets thing The spatial encoding data of part.That is, the spatial encoding data in internal memory can have both been searched according to memory headroom numbering, it is also possible to according to The numbering of each coding searches the spatial encoding data in internal memory in space encoding, can be combined with memory headroom numbering and space The numbering of each coding searches the spatial encoding data in internal memory in coding.Such as, two classes are combined for the third and numbers lookup Mode, in actual applications, specifically can according to memory headroom number order be positioned at sequence it is preceding storage first connect First data group with same projection direction of continuous memory headroom, then successively according to regional code numbering, ring difference coding numbering, Time-bin codings numbering and axial direction coding numbering position the spatial encoding data in first data group one by one in order, certainly Also other combinations, this is no longer going to repeat them.

Illustrate：

Referring to Fig. 5, according to the ranking results for meeting event space coded data, obtain meet event space coded number first According to 11111, before being carried out with reference to TOF information and data 11111 to and backwards projection calculate with complete this calculating, then obtain symbol Close event space coded data 11112, and whether compare data 11112 identical with the space encoding information of data 11111, if phase Together, then directly using the result of calculation of data 11111 as data 11112 result of calculation, if it is different, then in last computation result On the basis of, before being carried out with reference to TOF information and data 11112 to and backwards projection calculate with complete this calculating, the like, Obtain again and meet event space coded data 11113, and make above-mentioned same treatment, until completing to meet event space to each The iterative calculation of coded data.

It can be seen that, the embodiment of the present invention is based on the forward projection and backwards projection meter of the orderly list-mode data of complete space Calculate, compared with the reconstruction mode based on sinogram data for projection and routine list-mode data, difference is：First, with sky Between coded data substitute original list-mode completely and meet event data；Second, meet event space coded data suitable by space φ, Bin, RingDiff, Timebin, Z_Paral of sequence this five entry index codings are constituted, can be directly to meeting event data Carry out space orientation；3rd, after having processed one and having met event, just continue with it is next meet event, and whether need Meet event to this specifically to be calculated, can judge by the spatial ordering of data set, i.e. judge next meeting Whether event is identical with the current event space position (corresponding spatial encoding data) that meets, if identical, only will need to currently accord with The result of calculation of conjunction event be assigned to it is next meet event, so as to avoid unnecessary computing repeatedly.The advantage of this operation It is to improve the speed rebuild and calculate, while when data volume is significantly increased, reconstruction time is held essentially constant.

PET image reconstruction method provided in an embodiment of the present invention, by determining each in list-mode data to meet event The corresponding spatial encoding data of data, and all spatial encoding datas are sorted, it is ensured that the space of list-mode data is complete Order, using the spatial encoding data of the orderly list-mode of complete space, directly can carry out space to meeting event data Positioning, so as to need to only calculate non-zero, not only saves computing resource, has also speeded up reconstruction speed.

Additionally, in order to contrast list-mode algorithm for reconstructing of the tradition based on time order subset, tradition based on spacial ordering The method for reconstructing that the list-mode algorithm for reconstructing and the embodiment of the present invention of subset are proposed is rebuilding the difference in speed, respectively It is 1 × 10 to meeting event count⁷It is individual, 4 × 10⁷It is individual and 7 × 10⁷Individual list-mode data sets carry out reconstruction contrast.Wherein, Three kinds of methods all combine 22 OSEM algorithms of data subset, and carry out TOF reconstructions to same water mould scan data, by once After iterative calculation, the image that resolution ratio is for 256 × 256 × 65 is obtained, the reconstruction time of each method for reconstructing is as shown in the table：

(unit is contrasted using three kinds of TOF algorithm for reconstructing reconstruction times of different pieces of information amount：Second)

(threads of Core i7-3770 tetra-, 16GB RAM)

It can be seen that, compared with existing method for reconstructing, the reconstruction speed of embodiment of the present invention method for reconstructing is apparently higher than existing figure As method for reconstructing.

Referring to Fig. 7, the embodiment of the present invention additionally provides the structural representation of PET image reconstruction device, and the device 700 is wrapped Include：

Coding determining unit 701, for according to the table schema list-mode data rebuild for PET, determining respectively described Each in list-mode data to meet the corresponding spatial encoding data of event data, the spatial encoding data includes that first compiles Code, the second coding, the 3rd coding, the 4th coding and the 5th coding；

Event ordering unit 702, for all spatial encoding datas to be ranked up according to pre-set space order；

Image reconstruction unit 703, thing is met for the ranking results according to the event ordering unit 702 to each successively Part spatial encoding data is processed, and PET image reconstruction is completed with by iterative reconstruction algorithm；

Wherein,

Further, described device 700 also includes：

Direction encoding unit, for it is described coding determining unit 701 according to the list-mode data rebuild for PET, Determine respectively in the list-mode data it is each meet the corresponding space encoding of event data before, if each detector rings bag Containing M probe unit, then will meet projecting direction of the event in the fault plane and be divided intoPlant or M kinds, and to every kind of Projecting direction is encoded to obtainOr M direction encoding；

Regional code unit, for being divided into when projecting directionWhen planting, by same projection side in the fault plane M-1 downward view field is encoded, to obtain M-1 regional code, when projecting direction is divided into M kinds, by institute State in fault plane under same projection directionIt is individual orIndividual view field is encoded, to obtainIt is individual or Individual regional code；

Ring difference coding unit, if in the presence of Q detector rings being sequentially placed side by side, in fault plane axial direction On, the Q kind detector rings differences under same projection direction are encoded, to obtain P ring difference coding, Q≤P；

Wherein, M,Q, P, N, Z are natural number.

Wherein, the event ordering unit 702, including：

Wherein, described image reconstruction unit 703, including：

Or, described image reconstruction unit, including：

Further, described device 700 also includes：

And/or,

The event data obtains subelement, specifically for being numbered and/or to the space encoding according to memory headroom Numbering, each spatial encoding data for meeting event in memory headroom is obtained according to number order successively.

PET image reconstruction device provided in an embodiment of the present invention, by determining each in list-mode data to meet event The corresponding spatial encoding data of data, all spatial encoding datas for meeting event is ranked up, it is ensured that list-mode numbers According to the complete ordering property in space, using the orderly list-mode spatial encoding datas of complete space, can be directly to meeting event number According to space orientation is carried out, so as to need to only calculate non-zero, computing resource is not only saved, also speeded up reconstruction speed Degree.

As seen through the above description of the embodiments, those skilled in the art can be understood that above-mentioned implementation All or part of step in example method can add the mode of required general hardware platform to realize by software.Based on such Understand, the part that technical scheme substantially contributes to prior art in other words can be in the form of software product Embody, the computer software product can be stored in storage medium, such as ROM/RAM, magnetic disc, CD, including it is some Instruction is used to so that a computer equipment (can be the network services such as personal computer, server, or WMG Equipment, etc.) perform method described in some parts of each embodiment of the invention or embodiment.

It should be noted that for device disclosed in embodiment, because it is corresponded to the method disclosed in Example, So description is fairly simple, related part is referring to method part illustration.

Also, it should be noted that herein, such as first and second or the like relational terms are used merely to one Entity or operation make a distinction with another entity or operation, and between not necessarily requiring or implying these entities or operate There is any this actual relation or order.And, term " including ", "comprising" or its any other variant be intended to contain Lid nonexcludability is included, so that process, method, article or equipment including a series of key elements not only will including those Element, but also other key elements including being not expressly set out, or also include being this process, method, article or equipment Intrinsic key element.In the absence of more restrictions, the key element limited by sentence "including a ...", it is not excluded that Also there is other identical element in process, method, article or equipment including the key element.

The foregoing description of the disclosed embodiments, enables professional and technical personnel in the field to realize or uses the present invention. Various modifications to these embodiments will be apparent for those skilled in the art, as defined herein General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, the present invention The embodiments shown herein is not intended to be limited to, and is to fit to and principles disclosed herein and features of novelty phase one The scope most wide for causing.

Claims

1. a kind of PET image reconstruction method, it is characterised in that including：

According to the table schema list-mode data rebuild for PET, each in the list-mode data meeting is determined respectively The corresponding spatial encoding data of event data, the spatial encoding data includes the first coding, the second coding, the 3rd coding, the Four codings and the 5th coding；

All spatial encoding datas are ranked up according to pre-set space order；

Each spatial encoding data is processed successively according to ranking results, PET image is completed with by iterative reconstruction algorithm Rebuild；

Wherein,

Described first is encoded to and meets the corresponding direction encoding of projecting direction of the event in fault plane, and the fault plane is The annular section of detector rings；

Described second is encoded to and described meets the corresponding regional code of view field of the event under the projecting direction, the throwing Shadow zone domain is to detect the region met between two probe units of event；

Described 3rd be encoded to it is described meet the corresponding ring difference coding of detector rings difference that event has, the detector rings are poor Be the difference after number of rings with 1, it is described after number of rings be the view field on fault plane axial direction after spy Survey device number of rings amount；

Described 4th is encoded to the corresponding time-bin codings in time-bin positions for meeting event described in generation, the time- Bin positions be along the projecting direction in view field according to list-mode data time difference information divide each when Between interval in one；

Described 5th is encoded to the corresponding axial direction of axial location of the event in the time-bin positions that meets encodes.

2. method according to claim 1, it is characterised in that be used for the list-mode numbers that PET rebuilds in the basis According to, determine respectively in the list-mode data it is each meet the corresponding spatial encoding data of event data before, also include：

If each detector rings includes M probe unit, when thinking same probe unit in two neighboring probe unit When projecting direction between each probe unit is identical, projecting direction of the event in the fault plane will be met and be divided intoKind, when thinking that same probe unit is different to the projecting direction between each probe unit in two neighboring probe unit When, projecting direction of the event in the fault plane will be met and be divided into M kinds, wherein, the same probe unit with it is described A probe unit is spaced between two neighboring probe unit；And every kind of projecting direction is encoded to obtainIndividual or M Direction encoding；

When projecting direction is divided intoWhen planting, make the region between each two probe unit under every kind of projecting direction as one Individual view field, obtains M-1 view field under same projection direction in the fault plane, and to M-1 view field Encoded, to obtain M-1 regional code；

When projecting direction is divided into M kinds, the region between each two probe unit is made under some projecting directions as one Individual view field, makes the region between three probe units as a view field under some projecting directions, obtains described In fault plane under same projection directionIt is individual orIndividual view field, and it is rightIt is individual orIndividual view field enters Row coding, to obtainIt is individual orIndividual regional code；

If in the presence of P detector rings being sequentially placed side by side, on fault plane axial direction, by under same projection direction Q kind detector rings differences are encoded, to obtain Q ring difference coding, Q≤P；

Fault plane axial region is divided into the N number of time interval axially in parallel with fault plane, and to same projection direction Under N number of time interval encoded, encoded with obtaining N number of time-bin；

The Z kind axis projections region with same probe ring difference RD under to same projection direction carries out axial coding, with Axially encoded to Z, Z=P-RD；

Wherein, M,Q, P, N, Z are natural number.

3. method according to claim 1, it is characterised in that described that all spatial encoding datas are suitable according to pre-set space Sequence is ranked up, including：

Same first data group will be divided into all spatial encoding datas that should have equidirectional to encode, and by each described the One data group is ranked up according to the first preset order；

Same 3rd data will be divided into all spatial encoding datas that there should be identical ring difference coding in second data group Group, and each described 3rd data group is ranked up according to the 3rd preset order；

The same 4th will be divided into all spatial encoding datas that there should be identical time-bin coding in 3rd data group Data group, and each described 4th data group is ranked up according to the 4th preset order；

All spatial encoding datas in 4th data group are ranked up according to the 5th preset order.

4. method according to claim 3, it is characterised in that it is described according to ranking results successively to each space encoding number According to being processed, including：

Obtain each spatial encoding data for meeting event successively according to ranking results；If the current space encoding for meeting event Data are different from the spatial encoding data that upper meets event, then meet with reference to flight time TOF information and currently the sky of event Between coded data carry out before to and backwards projection calculate and currently meet event forward projection and backwards projection result obtaining；If working as Before meet event spatial encoding data it is identical with the spatial encoding data that upper meets event, then by upper one meet before event to Projection or backwards projection result are used as currently meeting event forward projection or backwards projection result；

Or,

Obtain each spatial encoding data for meeting event successively according to ranking results；For the current sky for meeting event for obtaining Between coded data, with reference to TOF information and currently meet event spatial encoding data carry out before to and backwards projection calculate to obtain Obtain and currently meet event forward projection and backwards projection result.

5. method according to claim 4, it is characterised in that methods described also includes：

Each described first data group is respectively stored into each the first contiguous memory space, and it is empty to each first contiguous memory Between be numbered according to first preset order；

Each described second data group is respectively stored into each the second contiguous memory space, and it is empty to each second contiguous memory Between be numbered according to second preset order；

Each described 3rd data group is respectively stored into each the 3rd contiguous memory space, and it is empty to each the 3rd contiguous memory Between be numbered according to the 3rd preset order；

Each described 4th data group is respectively stored into each the 4th contiguous memory space, and it is empty to each the 4th contiguous memory Between be numbered according to the 4th preset order；

Each spatial encoding data in 4th data group is respectively stored into each the 5th contiguous memory space, and to each Individual 5th contiguous memory space is numbered according to the 5th preset order；

And/or,

According to direction encoding each described first data group is respectively stored into each the first contiguous memory space, and to being related to All directions coding is numbered according to first preset order；

According to regional code each described second data group is respectively stored into each the second contiguous memory space, and to being related to Regional coding is numbered according to second preset order；

According to ring difference coding each described 3rd data group is respectively stored into each the 3rd contiguous memory space, and to being related to Each ring difference coding is numbered according to the 3rd preset order；

Each described 4th data group is respectively stored into each the 4th contiguous memory space according to time-bin codings, and to relating to And each time-bin coding be numbered according to the 4th preset order；

According to axially coding by each spatial encoding data in the 4th data group be respectively stored into each the 5th it is continuous in Deposit space, and to be related to each axially coding be numbered according to the 5th preset order；

Wherein, the second contiguous memory space belongs to the first contiguous memory space, the 3rd contiguous memory space category In the second contiguous memory space, the 4th contiguous memory space belongs to the 3rd contiguous memory space, the described 5th Contiguous memory space belongs to the 4th contiguous memory space；

Numbered according to memory headroom and/or to the numbering of space encoding, obtained successively according to number order every in memory headroom One spatial encoding data for meeting event.

6. a kind of PET image reconstruction device, it is characterised in that including：

Coding determining unit, for according to the table schema list-mode data rebuild for PET, the list- being determined respectively Each in mode data to meet the corresponding spatial encoding data of event data, the spatial encoding data includes the first coding, the Two codings, the 3rd coding, the 4th coding and the 5th coding；

Image reconstruction unit, is carried out to each spatial encoding data successively for the ranking results according to the event ordering unit Treatment, PET image reconstruction is completed with by iterative reconstruction algorithm；

Wherein,

7. device according to claim 6, it is characterised in that described device also includes：

Direction encoding unit, for, according to the list-mode data rebuild for PET, difference to be true in the coding determining unit In the fixed list-mode data it is each meet the corresponding spatial encoding data of event data before, if each detector rings includes M Individual probe unit, then, when thinking same probe unit to the projection between each probe unit in two neighboring probe unit When direction is identical, projecting direction of the event in the fault plane will be met and be divided intoKind, when thinking same probe unit To the projecting direction between each probe unit in two neighboring probe unit it is different when, event will be met and put down in the tomography Projecting direction in face is divided into M kinds, wherein, it is spaced one between the same probe unit and the two neighboring probe unit Individual probe unit；And every kind of projecting direction is encoded to obtainIndividual or M direction encoding；

Regional code unit, for being divided into when projecting directionWhen planting, each two under every kind of projecting direction is set to detect single Region between unit obtains M-1 view field under same projection direction in the fault plane as a view field, And M-1 view field is encoded, to obtain M-1 regional code；When projecting direction is divided into M kinds, at some Make the region between each two probe unit under projecting direction as a view field, three spies are made under some projecting directions The region surveyed between unit is obtained in the fault plane under same projection direction as a view fieldIt is individual orIndividual view field, and it is rightIt is individual orIndividual view field is encoded, to obtainIt is individual orCompile in individual region Code；

Ring difference coding unit, if in the presence of P detector rings being sequentially placed side by side, on fault plane axial direction, Q kind detector rings differences under same projection direction are encoded, to obtain Q ring difference coding, Q≤P；

Time encoding unit, for fault plane axial region to be divided into the N number of time interval axially in parallel with fault plane, And to same projection direction under N number of time interval encode, encoded with obtaining N number of time-bin；

Axial coding unit, for the Z kind axis projections region with same probe ring difference RD under to same projection direction Axial coding is carried out, to obtain Z axially coding, Z=P-RD；

Wherein, M,Q, P, N, Z are natural number.

8. device according to claim 6, it is characterised in that the event ordering unit, including：

First sequence subelement, for all spatial encoding datas that should have equidirectional to encode will be divided into same first number According to group, and each described first data group is ranked up according to the first preset order；

Second sequence subelement, for by first data group to should have same area encode all spatial encoding datas Same second data group is divided into, and each described second data group is ranked up according to the second preset order；

3rd sequence subelement, for by all spatial encoding datas in second data group to that should have identical ring difference coding Same 3rd data group is divided into, and each described 3rd data group is ranked up according to the 3rd preset order；

4th sequence subelement, for by all space encodings in the 3rd data group to that should have identical time-bin coding Data are divided into same 4th data group, and each described 4th data group is ranked up according to the 4th preset order；

5th sequence subelement, for all spatial encoding datas in the 4th data group to be entered according to the 5th preset order Row sequence.

9. device according to claim 8, it is characterised in that

Described image reconstruction unit, including：

Event data obtains subelement, for obtaining each spatial encoding data for meeting event successively according to ranking results；

First iterative calculation subelement, if meeting the space volume of event for the current spatial encoding data of event and upper that meets Code data are different, then with reference to flight time TOF information and currently meet event spatial encoding data carry out before to and inversely throwing Shadow calculating currently meets event forward projection and backwards projection result to obtain；

Iteration result assignment subelement, if meeting the space volume of event for the current spatial encoding data of event and upper that meets Code data are identical, then meet event forward projection or backwards projection result as currently meeting event forward projection or inverse using upper one To projection result；

Or, described image reconstruction unit, including：

Secondary iteration computation subunit, for for the current spatial encoding data for meeting event for obtaining, with reference to TOF information and The current spatial encoding data for meeting event carry out before to and backwards projection calculate with obtain it is current meet event forward projection and Backwards projection result.

10. device according to claim 9, it is characterised in that described device also includes：

First space numbered cell, for each described first data group to be respectively stored into each the first contiguous memory space, And each the first contiguous memory space is numbered according to first preset order；

Second space numbered cell, for each described second data group to be respectively stored into each the second contiguous memory space, And each the second contiguous memory space is numbered according to second preset order；

3rd space numbered cell, for each described 3rd data group to be respectively stored into each the 3rd contiguous memory space, And each the 3rd contiguous memory space is numbered according to the 3rd preset order；

4th space numbered cell, for each described 4th data group to be respectively stored into each the 4th contiguous memory space, And each the 4th contiguous memory space is numbered according to the 4th preset order；

5th space numbered cell, for by each spatial encoding data in the 4th data group be respectively stored into each Five contiguous memory spaces, and each the 5th contiguous memory space is numbered according to the 5th preset order；

And/or,

First coding numbered cell, for each described first data group to be respectively stored into each the first company according to direction encoding Continuous memory headroom, and all directions coding to being related to is numbered according to first preset order；

Second coding numbered cell, for each described second data group to be respectively stored into each the second company according to regional code Continuous memory headroom, and regional coding to being related to is numbered according to second preset order；

3rd coding numbered cell, for each described 3rd data group to be respectively stored into each the 3rd company according to ring difference coding Continuous memory headroom, and each ring difference coding to being related to is numbered according to the 3rd preset order；

4th coding numbered cell, for according to time-bin codings by each described 4th data group be respectively stored into each the Four contiguous memory spaces, and to be related to each time-bin coding be numbered according to the 4th preset order；

5th coding numbered cell, for distinguishing each spatial encoding data in the 4th data group according to axially coding Store each the 5th contiguous memory space, and to be related to each axially coding compiled according to the 5th preset order Number；

The event data obtains subelement, specifically for being numbered according to memory headroom and/or to the numbering of space encoding, according to Number order obtains each spatial encoding data for meeting event in memory headroom successively.