Embodiment
Here exemplary embodiment will be illustrated in detail, its example is illustrated in the accompanying drawings.Following description is related to
During accompanying drawing, unless otherwise indicated, the same numbers in different accompanying drawings represent same or analogous key element.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistent with the disclosure.On the contrary, they be only with it is such as appended
The example of the consistent apparatus and method of some aspects be described in detail in claims, the disclosure.
Fig. 1 illustrates the system architecture diagram of PET system, as shown in figure 1, PET system can include:Carrying subject is swept
Retouch bed 11, detector 12, detection process circuit 13 and image reconstruction device 14.Wherein, can be by sweeping when to tested swept-volume
Retouching the carrying of bed 11 drives injection to have the subject of radionuclide drug, is arrived along arrow direction shown in dotted lines in Figure 1
Detector 12, detector 12 can be received to the event of gammaphoton, referred to as single event in scanning.Detect process circuit 13
It can carry out meeting judgement according to the multiple single events received, obtain PET and meet event, and by the basis of image reconstruction device 14
Meet event reconstruction image.
Meet event to refer in a given small period in PET system, obtain two or more things
Part, this given period are referred to as meeting time window, generally several nanoseconds.It is that two events are referred to as two and met to meet event,
More than two events are referred to as multiple coincidence, generally only consider that two meet, and multiple coincidence can be carried out into discarding and be not involved in rebuilding.It is right
Meet event in two, if two events are produced by same positron annihilation and without scattering, referred to as true coincidence;If
Two events are to produce still at least one event by same positron annihilation to be scattered, and referred to as scattering meets;Such as
Two events of fruit are by different positron annihilation generations, referred to as random coincidence.All events that meet are properly termed as according with immediately
Data are closed, all true coincidence events are properly termed as true coincidence data, and all random coincidence events are properly termed as random coincidence data,
All scatterings, which meet event and are properly termed as scattering, meets data.
For image reconstruction device 14 when carrying out image reconstruction, the data being actually needed can be true coincidence data, and random
Meet data and scattering meets data and is properly termed as noise data, can be according to noise data to meeting data carries out school immediately
Just, to cause the quality of image reconstruction higher.For example, above-mentioned random coincidence data are a kind of main noise, can be in sight
Remove random coincidence data in Shi Fuhe data so that the data for participating in image reconstruction are more accurate.The image that the disclosure provides
It method for reconstructing, will mainly describe how to obtain accurate random coincidence data, and image weight carried out according to the random coincidence data
Build.
The example of the disclosure can use delay coincidence window method to obtain random coincidence data.As shown in figure 1, in PET system
Detection process circuit 13 in can include instant coincident circuit 131, delay coincidence circuit 132 and single event counting device 133.
Detecting process circuit 13 can be according to the subject scan data of collection, required for getting in random coincidence data calculation process
Parameter, for example, meeting data and delay coincidence data immediately.Image reconstruction device 14 can obtain according to detection process circuit 13
The parameter acquiring random coincidence data arrived, and carry out PET image reconstruction using random coincidence data.
Following example will be described during single pass, how be obtained random coincidence data and be carried out image reconstruction
Process.The processing procedure can be performed by image reconstruction device.Wherein, " true coincidence " number being related in follow-up description
According to can be that true coincidence data and scattering meet data sum, that is, " true coincidence " that is previously mentioned actually includes same positive electron
Bury in oblivion produce and without scattering event and by same positron annihilation produce but at least one event there occurs
The event of scattering.
Referring to the flow of the image rebuilding method of Fig. 2 examples, this method can be performed by image processing equipment 14.The image
Processing equipment 14 can include memory, processor, and store the computer that can be run on a memory and on a processor
Instruction, the computing device can realize following processing when instructing:
In step 201, detect process circuit and obtain and meet data, delay coincidence data and single event counting immediately.
In this step, detector 12 can receive gammaphoton, referred to as single event when scanning subject.Detection is handled
Circuit 13 can carry out meeting judging to obtain by instant coincident circuit 131 meets data immediately, meeting immediately in this example
Data can two meet data, can be represented with P.Single event counting device 133 can obtain single event counting, can use S
Represent.
Delay coincidence circuit 132 can be used for obtaining delay coincidence data.In this example, delay coincidence circuit 132 can be with
Only retain two using single channel windowing pattern and meet data.Delay coincidence window method as shown in Figure 3 meets schematic diagram, meet immediately and
Delay coincidence is all to judge that two meet event in two row sequences of events, wherein, " meeting immediately " shown in Fig. 3 is by i.e.
When coincident circuit obtain and meet data, and this meet data be two meet meet data, i.e., the symbol that two single events are formed
Conjunction event, and the multiple coincidence data that more than two single event is formed abandon.And delay coincidence is two row things in meeting immediately
Carry out meeting judgement after the wherein one row delay certain time (meeting time window for example, being much larger than) of part sequence, the thing delayed
Part, which arranges, is referred to as delay event column, and each event in the row is referred to as delay event, and another undelayed event column of row is referred to as instant
Event column, each event therein are referred to as instant event, when postponing between event column and instant event column still according to meeting
Between window carry out meeting judgement, obtain delay coincidence data, delay coincidence data can also two meet data.
Shown in Figure 3, single channel windowing pattern only can be opened a window by delay event or instant event, and described opens
Window is that the time span for meeting time window delimited by the event.Exemplified by this example can be by postponing event windowing, in Fig. 3
Round dot event be delay event (actual implement in, delay event can be by postponing tag identifier), inverted triangle event is i.e.
When event.For example, after the instant event 21 in instant event column is postponed into certain time, corresponding delay event can be obtained
22;In another example after instant event 23 is postponed into certain time, corresponding delay event 24 can be obtained.Assuming that to postpone event
24 open a window, and as schematically shown in Figure 3, arrow line are drawn by delay event 24 and represents to form one with delay event 24 as starting
Meet time window, meet if a single event falls into this and meets time window and then formed.Such as when being fallen into Fig. 3 in the presence of one
Between window instant event 25, then postpone event 24 and instant event 25 formation two and meet.
Delay coincidence circuit 132 can according to the method shown in above-mentioned Fig. 3, obtain delay event column and instant event column it
Between two meet data, as delay coincidence data, can be represented with R.
In step 202, according to data, delay coincidence data and single event counting is met immediately, true coincidence list thing is determined
Part affecting parameters, the true coincidence single event affecting parameters are used to represent to participate in prolonging by meeting the true coincidence single event in data immediately
Delay coincidence data deviation caused by meeting late.
In step 203, using true coincidence single event affecting parameters, the delay coincidence data collected are corrected,
And iteration corrected after random coincidence data.
In this example, event, i.e. delay coincidence data are met by the collection of delay coincidence circuit 132 two.But delay symbol
Close data and deviation be present, be not accurate random coincidence data.Due to event has been carried out into one in delay coincidence circuit 132
The delay fixed time so that true coincidence event is separated, and the judgement of delay coincidence, therefore delay coincidence are participated in independent single event
The single event that there may be in data in true coincidence event participates in, and it is not accurate random coincidence number to cause delay coincidence data
According to.This deviation be by true coincidence event single event participate in cause, if this partial data be corrected, will cause with
It is more accurate that machine meets data.
Single event in true coincidence event can be carried out to delay coincidence data by above-mentioned step 202 and step 203
(can be referred to as:True coincidence single event) introduce offset correction.True coincidence single event as described below participates in the several of delay coincidence
Situation:
Delay coincidence resolution principle according to exemplified by Fig. 3, delay coincidence are by " instant event column " and " delay event
Row " in wherein all the way event open a window, judge another way event whether fall within windowing formed meet in time window, fall into i.e.
Form delay coincidence.So, true coincidence single event participate in delay coincidence when, or can be used as open a window single event, or can
To be used as the single event fallen into.
Situation one:True coincidence single event opens a window;
The event to be opened a window in instant coincident circuit is equally the windowing event in delay coincidence circuit, that is, assumes a single thing
Part is the event of windowing in instant coincident circuit, then the single event is after delay coincidence circuit delay, corresponding delay thing
Part is also the windowing event of delay coincidence circuit.Described " windowing ", i.e., by the time of the single event delimit one meet
Time window, judge whether other single events fall into the time window and formed and meet, that is to say, that the time corresponding to the event of windowing is
Meet the starting time of time window.
For example:Fig. 3 example is referred to, in instant coincident circuit, real event part 21 forms two symbols with real event part 26
Close, and this two to meet opened a window by real event part 21, i.e., one delimited by the time of real event part 21 and meets time window (as schemed
Arrow line shown in 3 draws the line i.e. time window), above-mentioned real event part 26 falls into this and met in time window, with the shape of real event part 21
Meet into two.So, as described above, real event part 21 opens a window in instant coincident circuit, after real event part 21 is postponed, obtain pair
The delay event 22 answered, it will similarly be opened a window by the delay event 22.Similarly, single thing in the instant coincident circuit shown in Fig. 3
Part 23 is windowing event, although not having other single events in instant coincident circuit falls into time window, after the single event 23 delay
Postponing event 24, and windowing event corresponding to delay coincidence circuit.
The formation of true coincidence event, it is to be opened a window by real event part and real event part is fallen into instant coincident circuit, therefore, root
According to foregoing " event to be opened a window in instant coincident circuit is equally the windowing event in delay coincidence circuit ", then meet immediately
The real event part to be opened a window in circuit, its corresponding delay event can also open a window in delay coincidence circuit, but this delay event
It is the real event part in instant coincident circuit in fact, so referred to as real event part opens a window.For example, Fig. 4 illustrates the delay symbol of situation one
Conjunction forms state, postpones in event column by a true coincidence single event (can abbreviation real event part) windowing, in instant event column
One single event (not forming the event met, such as the single event 23 shown in Fig. 3) or real event part are (in instant coincident circuit
It is middle to form the single event met) fall into.This example can calculate this part and be opened a window by true coincidence single event the delay coincidence formed
Data.
, first can be according to step in order to which the real event part in computing relay event column opens a window the delay coincidence data to be formed
What is collected in 201 meets data P and single event counting S, divided by the sweep time of present scan subject immediately, obtains immediately
Meet data counts rate p and single event counting rate s.Wherein, single event counting rate s is segmented into sSAnd sT, sSRepresent separate single
Photon count rate, sTRepresent true coincidence single photon counting rate, s=sS+sT, sT=2t, t are true coincidence counting rates.If by this example
The actual random coincidence data counts rate that son is finally calculated is set as rReal, then following relation be present:
T=p-rReal;……………………..(1)
sS=s-2t;……………………..(2)
Wherein, above-mentioned formula (1) and formula (2) can participate in the iterative process in follow-up calculate, in the initial of iteration
R can be setRealEqual to r, r is the delay for obtaining delay coincidence data R divided by present scan subject sweep time
Meet data counts rate.
According to Poisson distribution probabilistic relation, the true coincidence single event in situation one can be obtained and opened a window the delay coincidence to be formed
The counting rate r of dataT.It should be noted that the method for this example in follow-up description, can first calculate the meter of related data
Digit rate, being finally multiplied by sweep time can obtain corresponding data.Such as the r in formula below (3)TActually true symbol
Close single event to open a window the data counts rate of the real event part to be formed windowing delayed data, if by the rTIt is multiplied with sweep time
Real event part windowing delayed data is obtained, but in order to describe simplicity, some follow-up places can be described as the windowing delay of real event part
Data rT。
rT=t* τ * (sS+t)*exp(-τ*(sS+t))……………………..(3)
When being corrected to delay coincidence data, delayed data that real event part can open a window subtracts from delay coincidence data
Go.
Situation two:True coincidence single event is fallen into;
In this situation, delay event column is opened a window by independent single event, still, falls into the instant event column for meeting time window
In event, it may be possible to independent single event, it is also possible to the single event in true coincidence event.Such as Fig. 5 a and Fig. 5 b example, its
In, Fig. 5 a signals are fallen into by the single event windowing in delay event column and the immediately single event in event column and form delay coincidence,
Fig. 5 b signals are fallen into by the single event windowing in delay event column and immediately the true coincidence single event in event column and form delay symbol
Close (can be any single event in true coincidence, with true representations of events in figure).
The delay coincidence data of Fig. 5 a and Fig. 5 b examples, can be the single event windowing delayed data actually got.But
Be, the single event windowing delayed data be not fully accurate random coincidence data, such as, the delay coincidence in Fig. 5 a be by
Two independent single events are formed, and are random coincidence data;But the delay coincidence in Fig. 5 b is by the real event in instant event column
Part falls into be formed, and is the error information that real event part participates in delay coincidence, is not real random coincidence data.
In addition, also a kind of delay coincidence is also to be fallen into by single event windowing and real event part, but it is true due to having simultaneously
Event and single event fall into form multiple coincidence, and cause to abandon the multiple coincidence data, so that will also be included in multiple coincidence two
The delay coincidence that single event is formed individual really is lost, data retrieving that this part is lost can be obtained it is more accurate completely
Random coincidence data.As Fig. 5 c examples delay coincidence in multiple coincidence, be by delay event column single event windowing and
Single event and real event part in instant event column are fallen into.
By foregoing description, it can be seen that, true coincidence single event falls into the delay coincidence data to be formed, and on the one hand may cause
It was not that random coincidence data meet (for example, shown in Fig. 5 b) originally to add some, on the other hand may also result in and lost
Some were that real random coincidence data meet (for example, shown in Fig. 5 c) originally, therefore, can be to being got in the situation two
Single event windowing delayed data be corrected.This example can be by the ginseng for being corrected to single event windowing delayed data
Number, which is referred to as effectively random zoom factor, the coefficient, can be expressed as β.Formula (4) as follows, actual random coincidence data
Can be that the delay coincidence data r got by delay coincidence circuit subtracts real event part windowing delayed data rT, and by effective
It is i.e. available that random zoom factor β zooms in and out correction.It is shown as with the counting ratemeter of these data:
rReal=β * (r-rT)…………………………(4)
The acquisition of effectively random zoom factor, can be by three kinds of delay coincidence situations shown in fig. 5 above a to Fig. 5 c
Data calculate respectively, are obtained in conjunction with three kinds of data.
, can be according to true coincidence counting rate t, separate single photon count rate s for Fig. 5 aS, closed using Poisson distribution probability
System, the delay coincidence data that independent single event is fallen into single event windowing delayed data are obtained, are properly termed as the first delay coincidence
Data.Shown in equation below (5):
rSS=sSO*τ*sS*exp(-τ*sS)*exp(-τ*t)…………………………(5)
Wherein, sSOThe counting rate of the single event windowing assumed that, τ is to meet time window.
, can be according to true coincidence counting rate t, separate single photon count rate s for Fig. 5 bS, closed using Poisson distribution probability
System, obtain the delay coincidence that true coincidence single event falls into and (has a true coincidence single event to fall into) in single event windowing delayed data
Data, the second delay coincidence data are properly termed as, shown in equation below (6):
, can be according to true coincidence counting rate t, separate single photon count rate s for Fig. 5 cS, closed using Poisson distribution probability
System, obtain to form the data of more delay coincidences because true coincidence single event and independent single event are each fallen within, be properly termed as the 3rd and prolong
Meet data late.Such case is can to form a random coincidence originally, but is met due to there is true coincidence single event to fall into
Event window, becomes multiple coincidence, so as to cause this due random coincidence event to be dropped.Shown in equation below (7):
rSTS=sSO*τ*sS*exp(-τ*sS)*[1-exp(-τ*t)]…………………(7)
According to above-mentioned formula (5) to formula (7), it is as follows to derive effectively random zoom factor:According to above-mentioned to the He of situation one
The explanation of situation two, the delay coincidence data r=r that delay coincidence circuit actual acquisition arrivesT+rSS+rST, and it is actual accurately random
It can be that the data of situation one will be subtracted in the delay coincidence data to meet data, and to being obtained after the Data correction of situation two,
rReal=rSS+rSTS。
According to r=rT+rSS+rSTAnd rReal=rSS+rSTS, it is also theorized that above-mentioned effective random zoom factor β meter
Calculate formula:
rReal/(r-rT)=(rSS+rSTS)/(rSS+rST);
So as to obtain:rReal=(rSS+rSTS)/(rSS+rST)*(r-rT)=β * (r-rT)。
Obtain above-mentioned formula (4):rReal=β * (r-rT);The real event part of situation one can be opened a window delayed data by
Subtracted in the delay coincidence data r of actual acquisition, obtain single event windowing delayed data r-rT, and by effectively scaling at random be
Number β zooms in and out correction to the single event windowing delayed data of the actual acquisition again, obtains random coincidence data rReal。
Wherein, effectively random zoom factor β=(r can also be obtainedSS+rSTS)/(rSS+rST)…………..(8)。
In formula (8), the first delay coincidence data r as denominatorSSWith the second delay coincidence data rSTSum, can
To be that the single event that arrives of actual acquisition opens a window delayed data, be by delay coincidence circuit it is actual get opened by single event
Window shape into delay coincidence data;And all it is by really according with as the second delay coincidence data of molecule and the 3rd delay coincidence data
Close the delay coincidence data that single event falls into time window formation, including the multiple coincidence that two collected meet and discarded, this portion
Divided data is properly termed as true delay coincidence data.Effectively random zoom factor β can be that true delay coincidence data are opened with single event
The ratio between window delayed data.
To sum up, r can be iterated to calculate according to following flowReal
The first step:Make rReal=r
Iteration
Wherein, effective random zoom factor β in above-mentioned iterative process is when calculating, due to rSS、rST、rSTSFormula in
All comprising single event windowing counting rate sSO, and the sSOThe unknown number assumed that, can be when calculating β according to β calculation formula
By parameter sSORemove, do not influence β calculating.By sSOAfter removing, former formula (5), (6), (7) can be changed to above-mentioned change
For the α in flowSS、αST、αSTS。
Usual iteration can obtain Exact Solutions three times, that is, try to achieve the counting rate r of accurate random coincidence datarea l, multiplied by
Upper sweep time obtains accurate random coincidence data total value RReal。
In addition, the above-mentioned example being corrected to delay coincidence data, be with and meanwhile consider situation one and situation two
Exemplified by delay coincidence formational situation, in other examples, can also bound fraction situation delay coincidence data are corrected.
For example can be corrected with according to circumstances one, or individually according to circumstances two be corrected, or can be combined with other symbols
Conjunction form is corrected.No matter which kind of mode, all it is according to true coincidence single event affecting parameters, corrects and prolong because real event part introduces
Meet the data error brought late.
In step 204, the scan data obtained according to the random coincidence data and the scanning subject is carried out
Image reconstruction.
For example, when carrying out image reconstruction using random coincidence data, two ways can be included.A kind of mode is basis
The random coincidence data, the scan data obtained to scanning the subject is corrected at random, and utilizes sweeping after correction
Retouch data and carry out image reconstruction;Another way can also carry out image using random coincidence data as the part for rebuilding data
Rebuild.
In this step, the random coincidence data that can be obtained to step 203 do smoothing processing, obtain low noise random coincidence
Data.Such as equation below can be utilized:
Wherein A and B is two continuous crystal chunks, as shown in fig. 6, any two crystal between A and B can be formed
Meet, RlkRandom counter between crystal l and k is represented,
Then the random coincidence data can be utilized, the work that undergoes reconstruction carries out image reconstruction.
The disclosure counts by using data, delay coincidence data, single event is met immediately, estimates real event part and participates in prolonging
Error existing for random coincidence data caused by meeting late, by obtaining true coincidence single event affecting parameters, and according to the parameter
Delay coincidence data are corrected, obtain accurate random coincidence data, so that the quality of reconstruction image is improved.
If the function of the data processing method of the disclosure is realized in the form of SFU software functional unit and is used as independent production
Product are sold or in use, can be stored in a computer read/write memory medium.Based on such understanding, the skill of the disclosure
The part or the part of the technical scheme that art scheme substantially contributes to prior art in other words can be with software products
Form embody, the computer software product is stored in a storage medium, including some instructions are causing one
Control and processing equipment perform all or part of step of each embodiment methods described of the disclosure.And foregoing storage medium bag
Include:USB flash disk, mobile hard disk, read-only storage (ROM, Read-Only Memory), random access memory (RAM, Random
Access Memory), magnetic disc or CD etc. are various can be with the medium of store program codes.
A kind of computer-readable recording medium that this example provides, it has the instruction that is stored thereon, when by one or
During multiple computing devices so that one or more processors configuration for executing data processing, this method include:It is tested to obtain scanning
What body obtained meets data, delay coincidence data and single event counting immediately;
Counted according to data, delay coincidence data and the single event of meeting immediately, determine that true coincidence single event influences ginseng
Number, the true coincidence single event affecting parameters are used to represent to participate in delay by the true coincidence single event immediately met in data
Delay coincidence data deviation caused by meeting;
Using the true coincidence single event affecting parameters, the delay coincidence data are corrected, and iteration obtains school
Random coincidence data after just;
Image reconstruction is carried out according to the scan data that the random coincidence data and scanning subject obtain.
The preferred embodiment of the disclosure is the foregoing is only, not limiting the disclosure, all essences in the disclosure
God any modification, equivalent substitution and improvements done etc., should be included within the scope of disclosure protection with principle.