Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.
PET is a compound that can participate in the blood flow or metabolic process of organism tissues and is labeled with a positron-emitting radionuclide (such as F-18, which may be referred to as a positive radionuclide) and is injected into an organism to allow the organism to perform PET imaging in the effective field range of PET. In the process, the positron nuclide in the tracer releases positron e +, and the released positron e + moves a certain distance in a living body and then is annihilated with negative electron e-in the surrounding environment to generate a pair of gamma photons with equal energy (511KeV) and opposite propagation directions (about 180 degrees). The detector of PET system can detect the gamma photon pair, analyze the existence of positive electron, reconstruct PET image reflecting the metabolism of organism tissue, obtain the concentration distribution of tracer in the tested organism, and the doctor can judge the focus of cancer and other diseases.
In order to obtain a clear image, the PET data obtained by scanning needs to be corrected before PET image reconstruction, and common data correction includes random correction, normalization correction, count loss correction, scattering correction, attenuation correction, and the like. The method provided by the present disclosure is also applied to the PET data correction stage before image reconstruction, and is used for eliminating ring artifacts of the reconstructed PET image. Since ringing artifacts are mainly due to the occurrence of pulse pile-up phenomena when detecting gamma photon pairs (since pulse pile-up may lead to missing counts on the one hand and to erroneous count positions on the other hand, ringing artifacts tend to occur in PET images), this disclosure refers to such correction that eliminates ringing artifacts as "local pulse pile-up correction". Referring to the example of fig. 1, it can be seen that the present disclosure requires the above-described local pulse pile-up correction during PET data correction prior to image reconstruction after acquisition of clinical data.
The method for removing ring artifacts in PET images of the present disclosure, which involves the service acquisition phase and the clinical scan phase of PET, is described as follows. In the service acquisition stage, mainly system calibration and correction processes, such as normalization correction, count loss correction, and the like, are performed. System-dependent calibration and correction factors are calculated from the correction data acquired during the service acquisition phase. In the clinical scanning stage, the clinical scanning data are calibrated and corrected by factors obtained in the service acquisition stage, and finally an ideal PET image is obtained. In particular, for the present disclosure, when performing local pulse pile-up correction on PET data, a "local pulse pile-up correction factor" will be used, and the determination of this factor requires the use of some parameters derived from the service acquisition phase. Therefore, in the following description, it will be described first how to determine the parameters to be used in the service acquisition phase, and then how to correct with the obtained local pulse pile-up correction factors in the clinical scanning phase.
And a service acquisition stage: at this stage, the pulse pile-up correction factor corresponding to each detector module Block needs to be determined, that is, each "Block pulse pile-up correction factor" is determined.
As shown in fig. 2, a detection arrangement for detecting gamma photon pairs in a PET system is schematically illustrated. As shown in fig. 2, the detection apparatus 200 of the PET system generally includes a plurality of detection rings 20 arranged along an axis, each detection ring 20 is assembled by a plurality of detector modules 21, which are "blocks" in the present disclosure. Each detector module 21 may be composed of a scintillation crystal and a photomultiplier, the scintillation crystal may absorb gamma photons and generate a certain amount of visible light photons according to energy of the gamma photons, and the photomultiplier converts a visible light signal generated by the scintillation crystal into an electrical signal for output, for example, into a pulse output. The above-described event of detecting the incidence of a gamma photon on a scintillation crystal can be referred to as a "single event".
For a single Block, the disclosure provides a "Block average factor model" for calculating an average factor Block avgfactor corresponding to each Block according to the model, and also provides a "Block average single event model" for calculating an average single event Block avgsingsingles corresponding to each Block according to the model. And calculating a 'Block pulse pile-up correction factor' corresponding to Block according to the two parameters of Block AvgFactor and Block AvgSingle.
Referring to the example of FIG. 3, the Block average factor model is built from normalized correction factors. As can also be seen from fig. 1, the PET system may perform normalization correction in advance before the local pulse pile-up correction, and during the normalization correction, calculate each normalization correction factor from the acquired normalization correction data, for example, each correction factor may be calculated using a CBN (Component-based normalization) factorial model shown in the following formula (1). The process of calculating the normalized correction factor from the normalized correction data may be performed in a conventional manner and will not be described in detail.
NCuivj=εuiεvjbubvcuimodDcvjmodDduvrkfuvguvr..
Wherein epsilonuiAnd εvjIs the detector crystal efficiency factor (including the crystal intrinsic efficiency part and the dead time influence part), buAnd bvIs axialBlock side factor, cuimodDAnd cvjmodDIs the cross-sectional block side factor, duvrkIs a crystal interference factor, fuvIs an axial geometric factor, guvrIs the radial geometry factor and D is the number of crystals in the cross section of the block detector. u and v respectively represent the rings in which the two detector crystals are located, i and j respectively represent the positions Of the two detector crystals corresponding to the LOR within the rings, r represents the LOR (Line Of Response) radial position, and k represents the relative position Of the LOR in the Block detector.
Instead, the present disclosure calculates the normalization correction factor at a certain count rate and builds a Block average factor model by using part of the normalization correction factor, i.e., the Block average factor model is built by part of the parameters used to calculate the normalization correction factor, e.g., the axial Block side factor and the cross-sectional Block side factor. For example, the axial Block side factor and the cross-sectional Block side factor corresponding to the crystal at the same position in all blocks may be summed and normalized using the axial Block side factor and the cross-sectional Block side factor obtained in the normalization correction, and the following formula (2) may be modeled:
.. equation (2)
Where a is the crystal axial position designation within the Block, t is the crystal cross-sectional position designation within the Block, M is the Block axial crystal number, D is the Block cross-sectional crystal number, and Norm is a normalization factor, which may be, for example, the Norm of all Block AvgFactorsa,tAveraging to ensure that the average value of BlockAvgFactor is 1; mod in equation (2) represents a modulo operation, e.g., 6mod5 ═ 1. In addition, in the above formula (2), the axial block flank factor is buFor example, the cross-section block side factor uses cuimodDHowever, the embodiment is not limited to this, and for example, b may be usedvReplacement buAnd use in combination of cvjmodDReplacement cuimodD。
As described above, the Block average factor model is established, and the parameter Block avgfactor of each Block can be obtained according to the model.
Fig. 4 illustrates the establishment of a Block mean single event model, which is derived from the count loss correction data obtained during the service acquisition phase.
For example, count loss correction data at multiple count rates may be collected by a high activity decay source model, and a Block mean single event model may be built from the count loss correction data. The count loss correction data may be directly acquired single-event data (the single event refers to a single photon count received by the crystal), or may be single-event data obtained by performing an inverse process on the coincidence event. The Block average single event model comprises a plurality of groups of factors, each group of factors is obtained at the same counting rate (same scanning), namely crystal counts at the same position in all blocks are summed and finally normalized, and the following formula (3) is the Block average single event model:
..
Wherein, a is the crystal in Block axial position sign, t is the crystal in Block cross section position sign, c is the count rate, M is Block axial crystal figure, D is Block cross section crystal figure, Norm is the normalization coefficient, guarantees that Block avgsingSingle mean value is 1, SuiThe system directly collects single event data or single event data obtained according to an event inverse process.
According to the formula (2) and the formula (3), two parameters, namely, Block AvgFactor and Block AvgSingle, can be calculated, and on the basis, the Block AvgFactor and the Block AvgSingle can be calculated according to a certain functional relationship to obtain a Block pulse pile-up correction factor. As shown in the following equation (4):
.. formula (4)
Wherein,is the Block pulse pile-up correction factor at a certain count rate c, and F is a functional relationship. For example, the Block pulse pile-up correction factor may be a ratio of a Block averaging factor model and a Block averaging single event model, e.g., F (a, B) ═ a/B.
In addition, it should be noted that in the example of the present disclosure, the calculation of the formula (4) may be set to be completed in the service acquisition phase, then, in the service acquisition phase, a Block pulse pile-up correction factor corresponding to each Block in the detector is obtained, and then, in the clinical scanning phase, the Block pulse pile-up correction factor is directly searched and used. Optionally, the calculation of formula (4) may be performed in the clinical scanning stage, and may be flexibly set in implementation.
And (3) a clinical scanning stage: at this stage, two BLOCK corresponding to each line of line LOR are determined, and according to the "BLOCK pulse pile-up correction factors" of the two BLOCK, a local pulse pile-up correction factor corresponding to the LOR is obtained, and the LOR is corrected by the factor. Wherein, the response line LOR: the tracer generates positron annihilation in a living body, simultaneously generates a pair of gamma photons which form 180 degrees with each other, the pair of gamma photons is received by two crystals at the same time, the two crystals can determine a line, namely a line of response (LOR), when the two crystals receive the gamma photons at the same time, the positron annihilation is determined to occur on the line, and the annihilation frequency on the line is increased once.
Fig. 5 shows the relationship between a Line of Response (LOR) and the above blocks, and as shown in fig. 5, a plurality of detection rings 20 form an internal space, and a biological body 30 is shown in the internal space formed by the detection rings, and a pair of gamma photons 32 generated by a positron annihilation event 31 occurring in the internal space are incident on a pair of detection Block modules Block21 in opposite directions and are detected by the pair of detection Block modules 21, so that the connecting Line between the pair of blocks can be referred to as LOR. PET data acquired during a clinical scan records each line of coincidence LOR detected, as in the example of FIG. 6, with a plurality of LORs 50.
In the example of the present disclosure, when performing local pulse pile-up correction on clinical data, i.e. each LOR is to be corrected, fig. 7 illustrates a flow of the method for removing ring artifacts in a PET image of the present disclosure, which may be applied to each LOR:
in step 701, for each LOR, two coincidence detector modules BLOCK corresponding to the LOR are determined. When the PET system records the LOR, the LOR can be recorded by a connecting line between which two blocks, so that the two blocks corresponding to the LOR can be easily obtained.
After obtaining two blocks corresponding to the LOR, in step 702, for each Block, a Block pulse pile-up correction factor Block pileup factor corresponding to the Block is obtained. More specifically, the LOR corresponds to two crystals, the two crystals belong to two blocks respectively, and when the Block pulse pile-up correction factor Block pileup factor is acquired, the count rate (average single-event count rate) of the Block where the crystal is located is determined according to the position of the crystal.
Combining with the formula (4), for example, the Block pulse pile-up correction factor, BlockPileupFactor, may be calculated according to blockavgsingfactor and BlockAvgFactor, where, according to the formula (2) and the formula (3), each crystal in Block corresponds to a factor, and BlockAvgFactor may be independent of the count rate, and may be a factor corresponding to a "crystal corresponding to LOR"; and the blockavgsingsingle can search the blockavgsingsingle corresponding to the counting rate c according to the counting rate c of the Block, namely, factors corresponding to the two factors of the counting rate c of the Block where the crystal + crystal corresponding to the LOR are located. After obtaining the blockavbsgsingle and blockavbagfactor, the two factors may be calculated according to a certain function to obtain a blockackpileupfactor, for example, a ratio of blockacksgsingle/blockackavgfactor may be used as the blockackpileupfactor.
In step 703, after the Block pulse pile-up correction factors Block pileup factor of the two blocks corresponding to the LOR are obtained, the local pulse pile-up correction factor corresponding to the LOR may be obtained according to the Block pulse pile-up correction factors Block pileup factor of the two blocks. For example, it can be calculated according to the following formula (5):
.. formula (5)
Wherein, CAAnd CBIs the count rate of two blocks corresponding to the LOR, LORPileupfactoruivjIs the local pulse pile-up correction factor for that LOR.
In step 704, the LOR is locally pulse pile-up corrected based on the local pulse pile-up correction factor obtained in step 703. For example, in the correction, the data recorded by the LOR may be multiplied by a local pulse pile-up correction factor.
According to the method for removing the ring artifact of the PET image, the Block pulse accumulation correction factor is calculated for the Block corresponding to the LOR, the local pulse accumulation correction factor corresponding to the LOR is further obtained, the LOR is corrected, the problem of pulse accumulation can be solved, the ring artifact of the PET image is removed, and the quality of the PET image is improved; in addition, in the calculation of the correction factors, the complexity and the calculation time caused by the modeling of the normalization factors under a plurality of counting rates are avoided by modeling only part of the normalization correction factors under a certain counting rate; by establishing a functional relationship between the Block average factor model and the Block average single event model, the problem that after the normalization correction factor is updated, the partial pulse accumulation processing is not matched with the normalization correction processing after updating can be avoided, and the normalization correction can be updated regularly during service.
After the PET data are corrected by using the method disclosed by the invention, the effect is better through experimental verification, as shown in FIG. 8, the left side of the image in FIG. 8 is before correction, and the right side is after correction, so that the effect of removing the ring-shaped artifact of the PET image is better as can be obviously seen from the comparison images before and after correction.
A PET (or PET/CT) scanning device for scanning PET images may be a PET system composed of a plurality of devices such as a scanning gantry, a table, a computer system, and an operation console. Wherein, the inside of the scanning frame is provided with a detector ring for scanning. The PET data correction for the removal of ring artifacts in PET images of the present disclosure is a data processing stage after the data acquisition for scanning, and may be performed by data processing software installed in a computer system, for example. As shown in fig. 9, the method of the present disclosure may be performed by a control device 91, and the control device 91 may include a processor 910, a communication interface 920, a memory 930, and a bus 940. The processor 910, the communication interface 920, and the memory 930 communicate with each other via a bus 940.
The memory 930 may store a logic instruction for removing the ring artifact of the PET image, and may be a non-volatile memory (non-volatile memory), for example. Processor 910 may invoke the remove logic instructions executing the PET image ringing artifact in memory 930 to perform the PET image ringing artifact removal method described above.
The function of the logic instruction for removing the ring artifacts in the PET image can be stored in a computer readable storage medium if it is implemented in the form of a software functional unit and sold or used as a separate product. Based on such understanding, the technical solutions of the present disclosure may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the methods according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above-mentioned logic command for removing the PET image ringing artifact may be referred to as "a device for removing the PET image ringing artifact", and the device may be divided into functional blocks. As shown in fig. 10, the apparatus may include: BLOCK determination module 1001, BLOCK factor determination module 1002, LOR factor determination module 1003, and LOR correction module 1004.
A BLOCK determination module 1001, configured to determine, for the LOR, two coincidence detector modules BLOCK corresponding to the LOR;
a BLOCK factor determining module 1002, configured to, for each BLOCK, obtain, according to a count rate of the BLOCK, a BLOCK pulse pile-up correction factor corresponding to the BLOCK at the count rate;
a LOR factor determining module 1003, configured to multiply the BLOCK pulse pile-up correction factors of the two BLOCKs to obtain a local pulse pile-up correction factor corresponding to the LOR;
and an LOR correction module 1004 configured to perform local pulse pile-up correction on the LOR according to the local pulse pile-up correction factor.
Further, the BLOCK factor determining module 1002 is configured to calculate a BLOCK pulse pile-up correction factor according to a BLOCK average factor model and a BLOCK average single event model corresponding to the BLOCK.
As shown in fig. 11, BLOCK factor determination module 1002 may include: an average factor sub-module 1101, an average single event sub-module 1102, and a synthesis processing sub-module 1103.
The average factor sub-module 1101 is configured to establish a BLOCK average factor model according to the axial BLOCK side factor and the cross-sectional BLOCK side factor in the normalized correction factor at a count rate.
The average single event submodule 1102 is configured to establish a BLOCK average single event model according to the count loss correction data at multiple count rates.
And the comprehensive processing sub-module 1103 is configured to calculate the BLOCK pulse pile-up correction factor according to the BLOCK average factor model and the BLOCK average single event model.
For example, the comprehensive processing sub-module 1103 is configured to use a ratio of the BLOCK average factor model and the BLOCK average single-event model as a BLOCK pulse pile-up correction factor.
Further, counting the loss correction data, comprising: the collected single event data or the single event data obtained by carrying out the inverse process on the coincidence events.
According to the PET image ring artifact removing device in the disclosed example, the Block pulse accumulation correction factor is calculated through the Block corresponding to the LOR, then the local pulse accumulation correction factor corresponding to the LOR is obtained, the LOR is corrected, the problem of pulse accumulation can be solved, the ring artifact of the PET image is removed, and the quality of the PET image is improved.
The above description is only exemplary of the present application and should not be taken as limiting the present application, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the scope of protection of the present application.