Continuous bed motion acquisition with axial short phantom for pet imaging system setup and quality control

Abstract

An improved method for time alignment (TA) procedure and crystal efficiency (CE) normalization estimation procedure for a PET scanner system is disclosed. In the TA procedure, modeled time of flight (TOF) data is compared to measured TOF data from an axial short cylinder phantom in order to find the time offset (TO) for each detector. The TO is then simultaneously estimated by matching the TOF centroids between the modeled and measured TOF data. In the CE estimation, a TOF reconstruction of CBM data on the axial short cylinder phantom is performed. Alternation between TOF image reconstruction and CE update eventually leads to correct estimation of activity and CE components.

Description

[0001] Cross-reference to related applications This application claims priority to U.S. Provisional Application No. 62 / 912,817, filed October 9, 2019, pursuant to 35 USC § 119(e), the entire contents of which are incorporated herein by reference. Technical Field

[0002] This disclosure generally relates to medical imaging, and more specifically, to systems and methods for time alignment calibration and crystal efficiency normalization for continuous bed motion acquisition modes in positron emission tomography (PET) systems. Background Technology

[0003] PET is a nuclear medicine modality used to image metabolic processes using gamma photons emitted by a radiopharmaceutical administered to the patient. Multiple PET data points are acquired in multiple directions to generate / reconstruct three-dimensional PET images and / or multiple slices of the PET images. Before image reconstruction, the raw PET image data is in projection / sine curve space. PET scans typically provide useful information about the functional status of body tissues and systems, such as the cardiovascular, respiratory, and / or other systems. PET scans are useful for indicating the presence of soft tissue tumors or reduced blood flow to certain organs or regions of the body.

[0004] Time alignment (TA) refers to the process of calibrating a PET scanner to make any remaining time offset between the two detectors consistent. Even with the hardware and electronics matched as carefully as possible, some time discrepancies may remain. These time differences degrade the performance of a time-of-flight (TOF) PET scanner because it detects the location of coincidence events along the line of response (LOR). To ensure the accuracy of the timing information in TOF data, TA procedures are performed on the TOF PET scanner periodically.

[0005] Time-of-Flight (TOF) data-reconstructed images exhibit low sensitivity to data inconsistencies, such as systematic errors in the correction factor. On the other hand, the quality of TOF reconstruction largely depends on the accuracy of the timing information. To ensure accurate timing information in TOF reconstructions in PET scanners, a timing calibration (TA) procedure, also known as timed calibration, is performed periodically in the PET scanner. An example of TA calibration for a PET scanner system is disclosed in U.S. Patent No. 8,089,043, issued January 3, 2012, the contents of which are incorporated herein by reference.

[0006] Typically, test subjects (models) are used to calibrate and / or validate (QC) the accuracy of nuclear medicine imaging equipment, such as PET scanners. A phantom is an object containing positron (β+) emission activity in a known shape and distributed throughout its body. Therefore, by imaging phantoms, the accuracy of PET instruments and their software can be evaluated, and settings can be adjusted if necessary.

[0007] Conventionally, cylindrical phantoms are used for quality control and calibration of PET, PET / CT, and PET / MR systems. These phantoms comprise 68Ge as a positron source, dispersed within a solid polyurethane matrix and encapsulated in a polyethylene shell. The cylindrical shape facilitates subsequent analysis through symmetry.

[0008] For the TA procedure, a uniform activity cylindrical phantom is placed at the center of the detector's field of view (FOV). It is assumed that all LORs originating from a particular detector have a TOF centroid centered at zero. Then, a time offset (TO) for each crystal is selected accordingly. The generated TOs are stored in the system and used during data reconstruction for each LOR TOF information shift, which is a combination of crystal TOs.

[0009] The cylindrical phantom is longer than the axial range of the imaging volume, so a size of 21 cm in diameter × 35 cm in length is typical. The phantom is quite heavy, weighing about 10-15 kg, and requires more than 100 kg of lead to protect technicians, scanners, patients, and others from the large amount of annihilated gamma photons when not in use.

[0010] In modern PET scanners, the scanner length has become longer in an attempt to extend the field of view (FOV) and thus gain sensitivity. Scaling phantoms to match the longer FOV of these PET scanners with their ever-increasing axial range is impractical due to excessive attenuation and scattering of gamma photons, and the phantoms become too heavy and difficult to shield during storage.

[0011] The development of detectors based on silicon photomultiplier tubes (SiPMs) has significantly improved the time-of-flight (TOF) resolution in PET scans. For example, Siemens' next-generation SiPM prototype PET / CT scanner has achieved a TOF resolution of approximately 250 ps. However, due to the improved TOF resolution in SiPM-based detectors, verifying TOF information requires higher standards, as TOF reconstruction is sensitive to any misalignment of events based on inaccurate time measurements.

[0012] Typically, PET scanner quality control (QC) procedures are performed daily on a uniform cylindrical phantom for PET scanner calibration, involving crystal efficiency (CE) estimation and time alignment (TA). However, such phantoms are relatively large and heavy, and not necessarily optimal for clinical settings. Recently, self-calibration methods have been introduced, such as patient-based QC procedures for TA and CE estimation on arbitrary subjects, including patients.

[0013] Therefore, an improved TA and CE estimation procedure is needed to provide more accurate system settings for the QC procedure of long FOV PET scanner systems. Summary of the Invention

[0014] This paper discloses a novel method for improving TA and CE estimation procedures for PET scanner systems with long FOVs by using a short axial calibration phantom with a shorter overall length compared to conventionally used calibration phantoms.

[0015] Continuous bed motion (CBM) acquisition and reconstruction is currently standard on many PET scanners. This type of acquisition opens up the possibility of using short and therefore lightweight phantoms for system setup. In CBM acquisition, counts from various detector pairs are combined, taking bed motion into account. While this is beneficial for accurate activity reconstruction, axial detector structure is lost due to averaging over motion. To make feasible estimates of detector characteristics, supplementary "static" data, similar to step-and-shoot (S&S) data, is generated. In this supplementary set, activity is integrated over motion while preserving the detector's axial structure. The same reconstructor can be used to generate both datasets; in one case, motion is integrated into the reconstruction process, and in the second case, it is ignored. Using these two supplementary datasets, simultaneous reconstruction of PET activity and PET detector characteristics can be achieved in CBM.

[0016] This paper discloses a TA setup procedure and crystal efficiency (CE) normalization estimation using an axially short uniform phantom during CBM system setup. While CBM “static” scans are naturally uniform in the axial direction, accurate modeling of these data requires reconstructing the CBM data. This disclosure relates to improvements in TA setup and CE estimation.

[0017] In the TA setup procedure, CBM data is generated without TOF information because the TO is unknown at this time. This data is then reconstructed using a standard non-TOF CBM reconstruction method.

[0018] The supplementary dataset acquired in the stationary coordinate system was modeled as a motion-blurred TOF projection from the previously step-reconstructed images. Note that attenuation is included in the blurring operation because it is not separated from the activity projection. The resulting TOF data is used in the TA procedure based on data momentum.

[0019] In some embodiments, a time alignment (TA) method for a PET scanner system is disclosed. The method includes: receiving list pattern data; generating non-time-of-flight (TOF) projection data, a scanner efficiency normalization array n, and an attenuation factor a; reconstructing an image f from the non-TOF projection data, correcting for scanner efficiency using the normalization array n, and correcting for attenuation using the attenuation factor a; and processing non-TOF and TOF scattering S and non-TOF and TOF projections. Modeling is performed; for each response line, the true TOF centroid (modeled TOFCOM) that conforms to the distribution is modeled from the zero-order moment M0 and first-order moment M1 of the modeled TOF data; for each response line, the measured TOF centroid (measured TOF COM) is calculated; and the TOF time offset is determined by subtracting the measured TOF COM from the modeled TOF COM. to .

[0020] In the CE estimation procedure, Time-of-Flight (TOF) reconstruction is performed on CBM data taken from an axial short phantom. The alternation between TOF image reconstruction and CE updates ultimately leads to accurate estimates of the activity and CE components until the scaling parameters are reached. CE estimates from the rotating rod acquisition (used to derive all normalized components) are used in the single image TOF reconstruction, followed by CE updates through ten iterations of the Maximum Likelihood Alternating Conditional Expectation (ML-ACE) algorithm described in VY, Panin, “Simultaneous activity and crystal efficiencies reconstruction: TOF patient-based detector quality control,” IEEE MIC 2014.

[0021] By using an axially shorter cylindrical phantom in the TA and CE estimation procedures, and by moving the short cylinder through a longer FOV, the axially longer FOV of the PET scanner is obtained.

[0022] In some embodiments, a non-transitory machine-readable storage medium encoded with program instructions for controlling a PET scanner is disclosed. When a processor executes the program instructions, the processor performs the methods disclosed herein for TA (Transient Attraction) and / or CE (Cut-Output) estimation for a PET scanner. Attached Figure Description

[0023] Figure 1 This is a high-level block diagram of a PET scanner system with a time alignment calibration manager according to an embodiment of the present disclosure; Figure 2 These are perspective views of a PET scanner detector and two-dimensional front views of a ring detector used with a phantom, according to embodiments of the present invention. Figure 3A This is a flowchart 300 of the TA procedure disclosed herein; Figure 3B This is a flowchart 400 of the TA procedure for a CBM-mode PET scanner disclosed herein; Figure 4A and 4B The diagram illustrates two coordinate systems in the CBM acquisition setup. Only the axial axes are labeled. Figure 5 This is a flowchart 600 illustrating a method for generating a CE normalized estimate for a PET scanner by scanning an axial short cylindrical phantom 60 using CBM scanning, according to some embodiments of the present disclosure. Figure 6 The dataset used in the steps of the ML-ACE algorithm is shown; Figure 7A An example of an axial short cylindrical body mold according to the present disclosure is shown, the axial short cylindrical body mold being placed on a standard bracket mounted on a CBM bed; Figure 7B The CBM data reconstruction is shown in the activity, cross-sectional (left), and coronal (right) views; Figure 7C The image shows CT-based attenuation maps, with cross-sectional (left) and coronal (right) views. Figures 8A-8D The modeling sine plot of the "static" scan data is shown, which is the motion-blurred attenuation projection of the axial short cylindrical phantom; Figures 9A-9C The detector crystal TO estimations performed using two different methods and their differences are shown. Figures 10A-10E The CE estimates are shown for both standard and CBM settings; Figure 11A Normalized difference images of a uniform cylinder reconstructed using a CBM axial short cylindrical phantom and a standard normalized phantom are shown. The top view is a cross-sectional view, and the bottom view is a coronal view. Figure 11B Normalized difference images of a uniform cylinder reconstructed using a virtual cylinder and reconstructed using standard normalization are shown. The top view is a cross-sectional view, and the bottom view is a coronal view. Figure 12It is applicable according to some embodiments Figure 1 A detailed block diagram of a computer system example. Detailed Implementation

[0024] This description of exemplary embodiments is intended to be read in conjunction with the accompanying drawings, which are to be considered part of the entire written description.

[0025] Figure 1 This is a high-level schematic diagram of a PET scanner system 100, which includes a PET detector ring 105, a control unit 101, and a patient bed 260. The control unit 101 may include a central controller 28, any necessary memory 26, a coincidence processor 110, a TA manager 135, and an image reconstruction unit 115. The TA manager 135 is a microprocessor appropriately configured to execute the TA procedures disclosed herein. The PET detector ring 105 includes a scintillator crystal 130 and a photomultiplier tube 125, provided in a ring configuration surrounding a patient tunnel in which the patient bed 260 moves.

[0026] Typically, prior to a PET scan, a short-lived radioactive tracer isotope is injected into a living subject, such as a patient (e.g., usually entering the bloodstream). A tracer isotope such as fluorodeoxyglucose (FDG) is a sugar. During the PET scan, data is recorded from the tissue containing the concentrated tracer as the tracer isotope decays.

[0027] As the tissue containing the concentrated tracer undergoes positron emission decay, it emits positrons, the antiparticles of electrons with opposite charges. These positrons eventually collide with electrons, producing a pair of annihilation (gamma) photons traveling in opposite directions. These gamma photons are detected when they reach the scintillator crystal 130 in the PET detector ring 105, creating a burst of light detected by the photomultiplier tube 125. This pair of photons travels in substantially opposite directions and is processed by the coincidence processing unit 110 to determine if the detected photon pair originated from a coincidence event. If so, a signal indicating the detected photon pair is sent to the image reconstruction unit 115 for use with image data generated using a mathematical image reconstruction procedure.

[0028] TA Setup Program The conformation processing unit 110 includes a TA manager 135, which facilitates any remaining TO calibration of the PET detector between two conforming detectors. Although the TA manager 135 is shown as part of the conformation processing unit 110, the TA manager 135 may be a separate, independent unit or part of another component of the PET system 100, such as the image reconstruction unit 115.

[0029] Figure 2This is a perspective view of a PET detector with a detector ring 105 according to an embodiment of the present invention. In this example, the detector ring 105 is part of the PET gantry 105A and has a radius of 427.6 mm and 672 crystals / gap elements (e.g., crystals A and B). The CBM patient bed 260 is located in front of the PET scanning gantry 105A. Figure 2 The right side is a schematic enlarged view of the detector ring 105, showing an example LOR 210 of the FOV 215 between detector crystal A and detector crystal B.

[0030] A TOF event is the time difference between the time it takes for a coincidence event to arrive at crystal A and the time it takes for a coincidence event to arrive at crystal B. Those skilled in the art will understand that the location of a TOF event can be calculated from the coincidence time difference. The location of a TOF event can be determined at a point on the LOR (Location of Time). When calculating the TO, the positive time direction is generally from detector crystal A to detector crystal B.

[0031] This document describes a TA procedure for calibrating the temporal alignment of a PET detector 105 using an axially short phantom. In the TA procedure described herein, a list pattern data file of the axially short phantom is available, and a CT scan of the axially short phantom is not required because the parameters of the axially short phantom are known, and the phantom's positioning can be determined from the PET data. However, in some embodiments, a CT scan of the axially short phantom may be performed.

[0032] The TA procedure in the CBM acquisition mode described herein is similar to the general self-TA procedure for using patient CBM scan data disclosed in U.S. Patent Application Publication No. 2020 / 0033491A1. The disclosure of U.S. Patent Application Publication No. 2020 / 0033491A1 is incorporated herein by reference. However, here, CBM scan data from an axially short phantom is used for accurate TA estimation.

[0033] TO (Crystal Time Offset) estimation Having space (radial ( ), azimuth ( ) and axial ( (including polar segment) projection index The TOF instantaneous dataset y with the TOF bin index T can be modeled by combining projection and scattering estimation S and the average random function using the following equation (1). The modeling projection is derived from the emitting object f, defined by the voxel index k, corrected for scanner efficiency by the normalized array n, and corrected for attenuation by a. The scattering estimate S is also corrected for scanner efficiency. in is the mean of the TOF hint dataset y, and C is the geometric projection system matrix. In non-TOF notation, the T index is omitted because summation is performed over T. By generally accepted convention, variables marked with a "‾" above them indicate the mean.

[0034] Figure 3A This is a flowchart 300 illustrating the TA procedure of this disclosure executed by TA manager 135. Referring to flowchart 300, in some embodiments, the system's TA manager receives list pattern data (box 305) from a PET scan of an axially short cylindrical phantom. The emission data used does not require a CT scan of the phantom. Based on the list pattern data, non-TOF projection data (box 310), as well as a normalized array n (box 312) and an attenuation factor a (box 314) are generated. Those skilled in the art of medical imaging should readily understand how these are generated from the list pattern data.

[0035] Flowchart 300 also covers another option for attenuation correction, where attenuation is obtained by CT scans of an axial short cylindrical phantom (box 307).

[0036] Next, image f (box 320) is reconstructed from the non-TOF projection data, corrected for scanner efficiency using a normalized array n, and corrected for attenuation using an attenuation factor a. In step 330, f is known, along with non-TOF and TOF scattering S and non-TOF and TOF projections. When performing modeling, it should be noted that, as should be easily understood by those skilled in the art of medical imaging, f, n, a, S, and It is a vector.

[0037] Then in step 350, for each LOR, the TOF data from the modeling is then processed. zeroth and first moments Modeling is performed to estimate the true Time-of-Flight centroid (COM) data that conforms to the model distribution. This is achieved using the following set of equations (2): ,in , and .

[0038] Note that normalization (elimination in the numerator and denominator) is not required in the COM estimate for each LOR. Note that M0 is independent of the time-resolution model. This means that M0 will remain constant for any scanner resolution, such as 250 or 500 ps. Furthermore, if the TOF curve in reality is not Gaussian but some other symmetric curve, M0 will still be the same value. Assuming a normalized symmetric TOF kernel, M1 is also independent of the time-resolution model in the real unscattered portion. This may be a significant advantage of the method of this invention, as uniform (across various LORs) time resolution cannot be guaranteed, and exact knowledge is not required.

[0039] In list mode processing, as indicated by step 340, the measured TOF COM is also calculated for each LOR using the following set of equations (3). : ,in It is the measured TOF COM, and .

[0040] Note that the total mass M0 in the denominator is calculated using modeled data instead of measured data to suppress noise amplification through division. LOR normalization (which is a normalization coefficient calculated separately for each LOR) is applied to the measured data. In previous notation methods, the sine curve was compressed data, consisting of projection bins containing more than one LOR. Therefore, normalization is combined across the LORs of the spectrogram data. The projected data in list mode is not compressed; each bin contains one LOR. The calculation of M1 involves subtracting the delayed event r and adding the instantaneous event y.

[0041] In step 360, the TOF time offset is determined. to This is achieved by measuring the TOF COM from equation set (3). With modeling TOF COM from equation set (2), Find the difference between To achieve this: .

[0042] Defined by a single detector crystal, identified by indices i and j, and TO is defined by... to Indication. An unweighted Gaussian model is used in the optimization procedure to estimate TO: Where L is the least squares difference between the measured time offset and the modeled time offset, and It is the LOR contribution factor, defined as follows: .

[0043] Note the contribution factor. Maintain the tracking order of i, j pairs because TOF information is directional. The following equation (5) is the update equation for the coordinate descent (CD) optimization algorithm: ,in (5).

[0044] In equation (5), l is the number of sub-iterations for sequential updates of a single crystal TO, and m is the number of iterations when all crystals are updated once. Note that equation (5) includes the fan sum of the data derived from crystal i, and N i Indicates the number of LORs that match between crystal i and all possible crystal j. Since TO is defined only as an additive constant, after each iteration of the algorithm, TO is factored by a constant factor. Updated to ensure features: (6).

[0045] CBM data acquisition for TA program The method used to determine TO is also applicable to CBM acquisition.

[0046] like Figure 4A and Figure 4B As shown, in CBM acquisition, there are two opposing coordinate systems. One is the room's coordinate system, aligned with the scanner. In the second case, the room and scanner are moving, while the patient is stationary. Scanner data is represented by axial plane coordinates z, which include polar angles. The moving object virtual LOR has the same cross-sectional coordinates but is represented by different axial coordinates. Description. We denote a function where we assign coordinates... Mapped to scanner system coordinates , where t is the acquisition duration. The function uses knowledge of the object / bed axial motion. The function aligns the axial coordinates of the room and the moving scanner system with the nearest neighbor approximation to maintain Poisson statistics during data reconstruction. The reconstructor produces two types of data: CBM for activity reconstruction and static data for TO estimation. Activity, i.e., PET images, is reconstructed from CBM non-TOF projection data. The supplementary dataset COM acquired in the stationary coordinate system is modeled using the following set of equations (7): (7) in, , and The above definition applies to static bed systems.

[0047] TOF data modified by fuzzy operator B To account for fuzziness during bed motion in CBM mode. The fuzziness operator B caused by CBM motion can be denoted as... (8) in, It is the decay correction constant for a specific isotope. This is the Kronecker notation, where d is the dead-time correction factor and t is the acquisition time. Note that the modeling data... The definition includes decay in fuzzy operations.

[0048] Once the motion fuzz modeling COM data is defined, the measured COM is calculated according to expression (3), now in the scanner coordinate system, and the TO estimation procedure is the same as described in (4)-(6).

[0049] Figure 3B This is a flowchart 400 illustrating the TA procedure of this disclosure in a CBM system executed by TA manager 135. Referring to flowchart 400, in some embodiments, the system's TA manager receives list pattern data from a PET scan of an axially short cylindrical phantom (box 405). The emission data used does not require a CT scan of the cylindrical phantom. Using the list pattern data, CBM non-TOF projection data is generated using CBM bed motion knowledge (box 411) (box 410). The bed motion data is stored in a list pattern data file and reported at fractions of a second intervals with an accuracy of fractions of a millimeter. This provides bed motion knowledge. CBM data generation is a dynamic process of assigning list pattern events to appropriate spectrogram planes. List pattern events converted into histograms in CBM spectrogram space utilize nearest-neighbor interpolation along the axial direction. In other words, events from the same detector pair are assigned to the next image data plane when the distance of CBM bed displacement is equal to the spacing between spectrogram planes. A normalized array n (box 412) is also generated using CBM bed motion knowledge (box 411). The normalized array n is calculated by simulating the movement of an object through the scanner, aided by monitoring basic scanner acquisition parameters such as single-photon count rate. The procedure is essentially a scanner normalization coefficient averaged specifically for each sine wave plane. Attenuation factor. a Also generated from list-pattern data (box 414). Those skilled in the art of medical imaging should easily understand how these are generated from list-pattern data.

[0050] Flowchart 400 also covers another option for attenuation correction, where the attenuation factor a Obtained from CT scans of an axial short phantom (box 407).

[0051] Next, image f (box 420) is reconstructed from CBM non-TOF projection data, corrected for scanner efficiency using a normalized array n, and adjusted using an attenuation factor. a Correction is performed for attenuation. In step 430, given f, the non-TOF and TOF scattering S of the motion blur and the non-TOF and TOF projection of the motion blur. Modeling is performed. The motion blur mentioned here is attributed to bed motion in the CBM model. It is important to note that, as should be readily understood by those skilled in the art of medical imaging, f, n, a, S, and It is a vector.

[0052] The process of reconstructing image f from the CBM non-TOF projection data mentioned in box 420 is known to those skilled in the art. An example of such a process can be found in: VY Panin et al., “Continuous Bed Motion on clinical scanner: design, data correction, and reconstruction,” Phys. Med. Biol. 59 (2014) 6153-6174, the entire contents of which are incorporated herein by reference.

[0053] Then in step 450, for each LOR, the TOF data from the modeling is then processed. zeroth moment and first moment Modeling is performed to estimate the true conformal distribution of the TOF centroid (COM) data collected in a stationary coordinate system. This is done using the system of equations (7) discussed above: ,in , ,and .

[0054] As discussed above regarding the static bed model calculations, note that normalization is not required in the COM estimates for each LOR (elimination in the numerator and denominator). Note that M0 is independent of the time-resolution model. This means that M0 will remain constant for any scanner resolution, such as 250 or 500 ps. Furthermore, if the actual TOF curve is not Gaussian but some other symmetric curve, M0 will still be the same value. Assuming a normalized symmetric TOF kernel, M1 is also independent of the time-resolution model in the true unscattered portion. This may be a significant advantage of the method of this invention, as uniform (across various LORs) time resolution cannot be guaranteed, and exact knowledge is not required.

[0055] In list mode processing, as indicated by step 440, the measured TOF COM is also calculated for each LOR using the following set of equations (3). : ,in It is the measured TOF COM, and .

[0056] Note that the total mass M0 in the denominator is calculated using modeling data instead of measurement data to suppress noise amplification through division. LOR normalization (which is a normalization coefficient calculated separately for each LOR) is used for the measurement data. In the previous notation, the sine curve was compressed data, which was a projection bin containing more than one LOR. Therefore, normalization is combined on the LORs of the spectrogram data. The list mode projection data is not compressed, and each bin contains one LOR. The calculation of M1 involves subtracting the delayed event r and adding the instantaneous event y.

[0057] In step 460, if in the case of a stationary bed, the TOF time offset is determined. to This is achieved first by measuring the TOF COM from expression (3). Modeling TOM COM from expression (7), Find the difference between To complete: .

[0058] Defined by a single detector crystal, identified by indices i and j, and TO is defined by using to The unweighted Gaussian model (4) discussed above is used in the optimization procedure to estimate TO.

[0059] Time Alignment The system TA program generates estimates stored in the system and applied at the coincidence processor level. This results in the list pattern file containing correct timing information for each coincidence photon pair (i.e., adjusted for TO). The system TA program is not executed frequently.

[0060] The TA procedure is the same as described above in the TO estimation and CBM acquisition scenarios, except that it estimates the TO for a specific dataset and considers the TO estimate during additional reassembly of the list-pattern data. In other words, the TA procedure detects the remaining TO with respect to the TO estimate stored on the system. Only the reassembled data will contain the correct timing information.

[0061] According to another aspect of this disclosure, a non-transitory machine-readable storage medium encoded with program instructions for controlling a PET scanner system 100 is disclosed. When a processor in the TA manager 135 of the PET scanner system executes the program instructions, the processor performs a method for TA for the PET scanner, said method comprising: (a) Receive list pattern data from PET scan of an axial short phantom in CBM mode; (b) Generate non-time-of-flight (TOF) projection data, scanner efficiency normalized array n, and attenuation factor. a ; (c) Reconstructing image f from non-TOF projection data, correcting for scanner efficiency using a normalized array n, and using an attenuation factor. a Correct for attenuation; (d) Non-TOF and TOF scattering S and non-TOF and TOF projection Modeling; (e) For each response line, the true TOF centroid (COM) that conforms to the distribution is estimated by modeling the zero-order moment M0 and the first-order moment M1 from the modeled TOF data; (f) For each response line, calculate the measured TOF COM; and (g) Determine the TOF time offset by taking the difference between the measured TOF COM and the modeled TOF COM. to .

[0062] In this disclosure, the terms “estimate” and “model” are used interchangeably to refer to the same thing when modeling is used to estimate certain parameters or values.

[0063] In some embodiments of non-transitory machine-readable storage media, step (e) is performed using the following expression: ; in , ;and ; in It indicates radial ( ),position( ) and axial ( The spatial projection index of the projection includes polar angle segments; where T is the TOF bin index; in It is estimated TOF COM data. It is the TOF data used for modeling. yes The first moment, yes The zeroth moment of , and C is the geometric projection system matrix.

[0064] In some embodiments of non-transitory machine-readable storage media, step (f) is performed using the following expression: ,in It is the measured TOF COM and .

[0065] According to another aspect, a non-transitory machine-readable storage medium encoded with program instructions for controlling a CBM PET scanner is disclosed. When a processor in the TA manager 135 of the CBM PET scanner system 100 executes the program instructions, the processor performs a method for time alignment of the CBM PET scanner, wherein the method includes: (a) Receive list pattern data from a PET scan of an axial short phantom in CBM mode; (b) Generate CBM non-time-of-flight (TOF) projection data, scanner efficiency normalized array n, and attenuation factor. a ; (c) Reconstruct image f from CBM non-TOF projection data, correct for scanner efficiency using a normalized array n, and apply an attenuation factor. a Correct for attenuation; (d) Non-TOF and TOF scattering of motion blur and non-TOF and TOF projection of motion blur Perform modeling; (e) For each response line, from the zeroth moment of the modeling TOF data and first moment The TOF centroid (COM) of the true coincident distribution collected in the stationary coordinate system is estimated. (f) For each response line, calculate the measured TOF COM; and (g) Determine the TOF time offset by taking the difference between the measured TOF COM and the modeled TOF COM. to .

[0066] In some embodiments of non-transitory machine-readable storage media, step (e) is performed using the following expression: ; in , ;and in It represents radial (ρ) and azimuth (ρ) ) and axial ( The spatial projection index of the projection includes polar angle segments; where T is the TOF bin index; in It is estimated TOF COM data. It is the TOF data used for modeling. yes The first moment, yes The zeroth moment of , and C is the geometric projection system matrix; and Furthermore, B is a fuzzy operator, defined as follows: .

[0067] In some embodiments of non-transitory machine-readable storage media, step (f) is performed using the following expression: ,in It is the measured TOM COM and .

[0068] CE normalized estimation According to another aspect, a method for scanner performance monitoring (i.e., scanner calibration) of CBM acquisition patterns using an axial short phantom is disclosed. Taking axial bed motion into account, CBM data from various detector pairs are formed and counted together. While axial CE structure is beneficial for activity reconstruction, it is actually lost in the corresponding calculated normalized array due to averaging of axial bed motion. Additional supplementary data is generated during CBM acquisition pattern scanning of the axial short phantom, as disclosed in the step-and-scan (S&S) static acquisition method disclosed in U.S. Patent Application Publication No. 2015 / 0297168.

[0069] In the supplementary dataset, the activity from the TOF data of the axial short phantom is integrated over the axial motion of the patient bed, while the CE structure is preserved. The same reconstructor can be used to generate both datasets. Using the two supplementary datasets, it becomes feasible to simultaneously reconstruct the activity and scintillation CE normalized components from the TOF data of the axial short phantom acquired from the CBM acquisition process. The algorithm can be interpreted as a conventional activity maximum likelihood (ML) reconstruction with nested loops of scintillation CE normalized component estimation, using the same data compressed in different ways.

[0070] By considering the axial short phantom TOF data integrated over CBM acquisition motion, the axial short phantom image acquired by CBM and the scintillation CE normalization coefficient can be estimated. The CE normalization estimation procedure discussed herein is similar to that disclosed in U.S. Patent No. 10,482,596, which is extended for use with patient scan data in CBM acquisition mode. The disclosure of U.S. Patent No. 10,482,596 is incorporated herein by reference. However, CBM scan data of an axial short phantom are used here to generate accurate CE normalization estimates.

[0071] Only the normalized component of the CE needs to be estimated. This further significantly reduces the number of unknowns and the sensitivity of the solution to noise, but requires the development of a specific reconstruction algorithm. ML reconstruction of the activity and crystal efficiency (A-CE) problem has an additional aspect: scattering event modeling is involved in efficiency estimation, and the model equations are nonlinear with respect to scintillation CEs with known activities. The decay and activity are defined as proportionality constants because only their product is used in the estimation. Therefore, global scaling of the unknowns is a smaller problem because the average efficiency value is known due to the separate calibration of voxel activity with the injection dose of the radioactive tracer.

[0072] The central controller 28 operates according to stored instructions to perform various actions described herein, such as determining decay correction efficiency, determining single-photon count rate at a given time, determining detection time efficiency, calculating normalization coefficients, normalizing response line data, and / or reconstructing data. The central controller 28 is configured by software and / or hardware to perform any or all of the actions of the methods disclosed herein.

[0073] Figure 5 This is a flowchart 600 illustrating a method for generating a CE normalized estimate for a PET scanner by scanning an axially short cylindrical phantom 60 via CBM, according to some embodiments of the present disclosure. The axially short cylindrical phantom 60 is mounted on a patient bed 206, which moves the phantom during CBM acquisition of the phantom 60, such as... Figure 7A As shown. The method includes (a) generating TOF data of an axially short cylindrical phantom 60 in a CBM acquisition mode, taking into account the axial motion of the patient bed 206 during CBM acquisition, as a first dataset (see box 610); (b) generating a supplementary dataset by integrating the TOF data of the axially short cylindrical phantom 60 over the axial motion of the patient bed 206 during CBM acquisition (see box 620); and (c) simultaneously reconstructing the activity and CE normalization coefficients from the supplementary dataset (see box 630).

[0074] The operation and configuration of the central controller 28 are described first below. An example implementation is described in more detail in the following discussion.

[0075] The central controller 28 is configured to determine normalization coefficients. Weights are calculated to account for differences in the efficiency of different response lines relative to the patient. In CBM, the axial orientation of the response line changes over time when considered relative to the patient. Consequently, various detector pairs and other factors contribute to the efficiency of detection. These factors, which may change over time, are included in the calculation of the normalization coefficients.

[0076] The central controller 28 takes decay into account. It determines decay correction efficiency for the isotopes used during PET scans. The decay characteristics of the isotopes change over time. This change in decay is used to determine the decay correction efficiency.

[0077] The central controller 28 takes into account changes in bed and patient speed. As the bed speed changes, the detection time efficiency also changes. The detection time efficiency at different times and corresponding locations on the bed or patient during a PET scan is determined in part based on speed.

[0078] The central controller 28 may also take into account other factors, such as the time variation of the scanner’s single-photon count rate and / or the scanner’s normalization (e.g., the normalization used in S&S or other scanning protocols, in which there is no axial bed movement during scanning).

[0079] Central controller 28 applies normalization coefficients. For each given LOR, the activity is weighted by the normalization coefficients of that LOR. Central controller 28 or another processor can reconstruct the object space based on the normalized LORs.

[0080] The central controller 28 processes data using events (e.g., response line events), empirical information (e.g., global single-photon count rate), and / or known information (e.g., decay correction constants) stored in memory 26. For processing, data may bypass memory 26, be temporarily stored in memory 26, or be loaded from memory 26.

[0081] Detected events, LOR information (e.g., sine waves), time steps, single-photon count rates, decay information, scanner normalization information, CBM normalization coefficients, reconstructed images, or other data are stored in memory 26. Data is stored in any format. Memory 26 is a buffer, cache, RAM, removable media, hard disk drive, magnetic, optical, database, or other memory now known or developed in the future. Memory 26 is a single device or a group of two or more devices. Memory 26 is part of the PET system 100 or a remote workstation or database, such as a PACS memory.

[0082] Memory 26 is additionally or alternatively a non-transitory computer-readable storage medium having processing instructions. Memory 26 stores data representing instructions executable by a programmed central controller 28 for calculating normalization coefficients in continuous bed motion acquisition. Instructions for implementing the processes, methods, and / or techniques discussed herein are provided on a non-transitory computer-readable storage medium or memory, such as a cache, buffer, RAM, removable media, hard disk drive, or other computer-readable storage medium. Computer-readable storage media include various types of volatile and non-volatile storage media. Functions, actions, or tasks illustrated in the figures or described herein are performed in response to one or more sets of instructions stored in or on a computer-readable storage medium. The functions, actions, or tasks are independent of a particular type of instruction set, storage medium, processor, or processing strategy and may be operated individually or in combination by software, hardware, integrated circuits, firmware, microcode, etc. Similarly, processing strategies may include multiprocessing, multitasking, parallel processing, etc. In one embodiment, the instructions are stored on a removable media device for reading by a local or remote system. In other embodiments, the instructions are stored at a remote location for transmission over a computer network or telephone line. In yet another embodiment, the instructions are stored within a given computer, CPU, GPU, or system.

[0083] PET system 100 may include a display. For example, a central controller 28 reconstructs the scanned patient or object from normalized response line data. This reconstruction is used for three-dimensional rendering, multiplanar reconstruction, or two-dimensional imaging of the object or patient's functions. The image is displayed on the display. The display can be a CRT, LCD, plasma screen, projector, printer, or other output device for displaying images.

[0084] CBM Data Restructuring According to one aspect of this disclosure, a method for monitoring scanner performance during QC scanning of an axially short phantom using CBM acquisition involves CBM data reconstruction. (As described above...) Figure 4A and 4B As described, two coordinate systems exist in CBM acquisition. One is the laboratory coordinate system, which is aligned with the scanner. The second is the moving coordinate system, where the object is stationary. CBM scanner data consists of cross-sectional radial coordinates. Azimuth coordinates The cross-sectional coordinates are represented by z-coordinates, including polar angles. The virtual LOR (Land of the Moving Object) has the same cross-sectional coordinates but is represented by different axis coordinates. Description. Let's denote a function where we will use coordinates... Mapped onto the scanner system coordinates, for This function utilizes knowledge of the object / bed axial motion. It approximates the axial coordinates of the laboratory and mobile systems with the nearest neighbor to preserve Poisson statistics during data reconstruction. The reconstructor produces two types of data: one dataset for A (activity reconstruction) and a supplementary dataset for CE estimation.

[0085] Activity Reconstruction Steps With spatial projection index The TOF instantaneous CBM data y of the TOF bin index T can be modeled by combining the following: modeling projection of real events (non-scattering gamma coincidence events) from the emitting object f. Defined by the voxel index k, with attenuation corrected for by a through a normalized array n for scanner efficiency; and scattering estimate S, also corrected for scanner efficiency; and average random data. : Where C is the geometric projection system matrix, which is contained in The probability of detecting a coincidence event originating from voxel k of the activity array, and This refers to crystal efficiency. CBM data normalization (from coordinates) The indicator is calculated from the scanner normalization factor (integrated over time) according to the following formula, taking into account axial movement during CBM: in, It is the decay correction constant for a specific isotope. It is a Kronecker notation, d is the dead-time correction factor, and t acq This is the CBM acquisition time. Dead time correction—which adapts to signal loss when the detector is handling relatively high count rates—is time-dependent because the single-photon count rate is variable. Equation (2) describes the averaging process of virtual LOR efficiency relative to scanner LOR efficiency and is discussed in more detail by the inventors in “Continuous bed motion on clinical scanner: Design, DataCorrection and Reconstruction” by VY Panin, AM Smith, J. Hu, F. Kehren and ME Casey (Phys. Med. Bio., vol. 59, pp. 6153–6174, 2014).

[0086] The scanner normalization factor (indicated by the coordinate z) can accommodate the mixing and recombination of LORs, as in Siemens scanners, which connect two detector crystals i and i' to the projection chamber: , Where g is the geometric component of the normalized array. The LOR contribution factor is due to the hybrid and cross-data compression techniques that introduce LOR into the projected data. Defined by the following formula: (4).

[0087] Due to the integration of axial movement in (2), the CBM normalized array loses information about the axial CE structure. Therefore, assuming the CE is known, the described dataset is used for activity reconstruction.

[0088] CE estimation steps (CE steps) The supplementary dataset, collected in a stationary coordinate system, is real-time data, and modeled as follows: Among them, motion fuzzy operator It can be marked as .

[0089] Note the realism of CE step modeling. The definition includes attenuation in fuzzy operations. Due to the integral on the movement in (6), the axial structure of the activity distribution will be lost.

[0090] The total acquisition time can differ between (2) and (6) because the modeling in (5) should not include projections outside the axial range of the reconstructed field of view (FOV). The random mean also differs between the two datasets and is estimated differently from the delayed events.

[0091] ML-ACE algorithm It can be used Figure 6 The two datasets shown were subjected to simultaneous activity and crystal efficiency normalized component reconstruction (A-CE). Figure 6 In the diagram, the first column represents the activity prediction modeled. The second column represents the normalized array, the direct plane. The first row is the activity reconstruction step (A step), and the second row is the CE normalized component reconstruction step (CE step). The vertical axis is the axial direction. Optimization is performed iteratively; each step is divided into the two steps mentioned above. The A step is the activity update using a fixed normalized (efficiency) array, where CBM TOF data is used. A commonly used ordered subset expectation-maximization (OS-EM) algorithm is employed. The objective function is as follows. To maximize f, where f is the activity graph as defined in (1).

[0092] The CE step is an efficiency update of the iterative algorithm described in VY Panin’s “Monotonic Iterative Algorithms for Crystal Efficiencies Estimation from Normalization Data and Single Rates Estimation from Compressed Random Coincidence Data” (2013 IEEE Nucl. Sci. Symp. and Med. Imag. Conf. (Seoul, Korea), M23-1, 2013), where the activity and scattering distribution of the motion blur are known from step A. In this step, supplementary TOF or non-TOF data are used, and four iterations are performed in the CE estimation. Here, the goal is to... Maximize the following objective function: .

[0093] For completeness, the CE step update equation is provided: Where m is the crystal renewal efficiency The number of iterations. Note that TOF information is eliminated in the calculations of A and B, where and Indicates the position of the TOF repository index T and The sum of T is calculated. However, the TOF information is preserved in the calculation of C. If non-TOF data is used in the CE step, the calculation of C does not sum over the T index.

[0094] Each step uses a synchronous monotonic update algorithm. However, these two steps together can represent a sequential update method. This method can be interpreted as a regular activity ML reconstruction with nested loops and CE normalized component estimates, which effectively uses the same data but compresses it in a different way. In the TOF / non-TOF nested loops below, ML-ACE will label the CE normalized component estimates from the TOF / non-TOF CE step data.

[0095] The initial conditions for ML-ACE are CE initiated by the average block value and uniform activity distribution. Three iterations of the sequential update method are performed. This effectively results in 3 iterations and 21 subsets of OS-EM activity reconstruction, and 12 iterations of CE normalized component estimation.

[0096] Figure 12An exemplary computing environment 900 is illustrated, which includes Figure 1 An embodiment of the central controller 28. For example, a computing environment 900 can be used to implement the methods disclosed herein. Computers and computing environments such as the central controller 28 and the computing environment 900 are known to those skilled in the art and are therefore briefly described herein.

[0097] like Figure 12 As shown, the central controller 28 may include communication mechanisms, such as a system bus 921 or other communication mechanisms for transmitting information within the central controller 28. The central controller 28 further includes one or more processors 920 coupled to the system bus 921 for processing information.

[0098] Processor 920 may include one or more central processing units (CPUs), graphics processing units (GPUs), or any other processor known in the art. More generally, the processor may include means for executing machine-readable instructions stored on a computer-readable medium for performing tasks, and may include any or a combination of hardware and firmware. The processor may also include memory storing machine-readable instructions executable for performing tasks. The processor acts on information by manipulating, analyzing, modifying, transforming, or transmitting information for use by an executable program or information device, and / or by routing information to an output device. For example, the processor may use or include the capabilities of a computer, controller, or microprocessor, and be modulated using executable instructions to perform special functions that a general-purpose computer does not perform. The processor may be coupled (electrically coupled and / or, if including, executable components) to enable interaction and / or communication between them. The user interface processor or generator may include electronic circuitry or software, or a combination of both, for generating display images or portions thereof. The user interface may include one or more display images that enable a user to interact with the processor or other devices.

[0099] Continue to refer to Figure 12The central controller 28 also includes a system memory 930 coupled to a system bus 921 for storing information and instructions to be executed by the processor 920. The system memory 930 may include computer-readable storage media in the form of volatile and / or non-volatile memory, such as read-only memory (ROM) 931 and / or random access memory (RAM) 932. RAM 932 may include one or more other dynamic storage devices (e.g., dynamic RAM, static RAM, and synchronous DRAM). ROM 931 may include one or more other static storage devices (e.g., programmable ROM, erasable PROM, and electrically erasable PROM). Furthermore, the system memory 930 may be used to store temporary variables or other intermediate information during instruction execution by the processor 920. A basic input / output system 933 (BIOS)—which contains basic routines such as those that facilitate the transfer of information between elements within the central controller 28 during startup—may be stored in the ROM 931. RAM 932 may contain data and / or program modules that are immediately accessible to the processor 920 and / or currently being operated by the processor 920. The system memory 930 may additionally include, for example, an operating system 934, application programs 935, other program modules 936, and program data 937.

[0100] The central controller 28 may also include a disk controller 940 coupled to the system bus 921 to control one or more storage devices, such as magnetic hard disks 941 and removable media drives 942 (e.g., floppy disk drives, compact disk drives, tape drives, and / or solid-state drives), for storing information and instructions. Storage devices can be added to the central controller 28 using appropriate device interfaces, such as Small Computer System Interface (SCSI), Integrated Device Electronics (IDE), Universal Serial Bus (USB), or FireWire.

[0101] The central controller 28 may also include a display controller 965 coupled to the system bus 921 to control a display or monitor 966, such as a liquid crystal display (LCD), for displaying information to the computer user. The computer system includes an input interface 960 and one or more input devices, such as a keyboard 962 and a pointing device 961, for interacting with the computer user and providing information to the processor 920. The pointing device 961 may be, for example, a mouse, light pen, trackball, or joystick, for transmitting directional information and command selection to the processor 920 and for controlling cursor movement on the display 966. The display 966 may provide a touchscreen interface that allows input to supplement or replace the directional information and command selection communication provided by the pointing device 961.

[0102] Central controller 28 may execute some or all of the processing steps of the embodiment by executing one or more sequences of one or more instructions contained in memory (such as system memory 930) in response to processor 920. Such instructions may be read into system memory 930 from another computer-readable medium, such as magnetic hard disk 941 or removable media drive 942. Magnetic hard disk 941 may contain one or more data stores and data files used by various embodiments. The data store contents and data files may be encrypted to improve security. Processor 920 may also be employed in a multiprocessing arrangement to execute one or more sequences of instructions contained in system memory 930. In alternative embodiments, hardwired circuitry may be used instead of or in combination with software instructions. Therefore, the embodiments are not limited to any particular combination of hardware circuitry and software.

[0103] Some embodiments include software instructions written in high-level languages ​​such as C, C++, C#, Java, Fortran, or Python. Some embodiments are written for multi-paradigm numerical computation environments, such as Matlab, sold by Mathworks Inc. in Natick, Massachusetts.

[0104] As described above, the central controller 28 may include at least one computer-readable medium or memory for storing instructions and for containing data structures, tables, records, or other data as described herein. As used herein, the term "computer-readable medium" refers to any non-transitory machine-readable storage medium that participates in providing instructions to the processor 920 for execution. Computer-readable media can take many forms, including but not limited to non-transitory, non-volatile, and volatile media. Non-limiting examples of non-volatile media include optical discs, solid-state drives, magnetic disks, and magneto-optical discs, such as magnetic hard disk 941 or removable media drive 942. Non-limiting examples of volatile media include dynamic memory, such as dynamic random access memory 930.

[0105] Central controller 28 can operate in a networked environment using a logical connection to one or more remote computers, such as remote computing device 980. Remote computing device 980 can be a personal computer (laptop or desktop), mobile device, server, router, network PC, peer-to-peer device, or other common network node, and typically includes many or all of the elements described above relative to central controller 28. When used in a networked environment, central controller 28 may include modem 972 for establishing communication over network 971 (such as the Internet). Modem 972 may be connected to system bus 921 via user network interface 970 or via another suitable mechanism.

[0106] Network 971 may include, but is not limited to, the Internet, intranet, local area network (LAN), wide area network (WAN), metropolitan area network (MAN), personal area network (PAN), direct connection or a series of connections, cellular telephone network, or any other network or medium capable of facilitating communication between central controller 28 and other computers (e.g., remote computing device 980). Network 971 may be wired, wireless, or a combination thereof. Wired connections may be implemented using Ethernet, Universal Serial Bus (USB), RJ-6, or any other wired connection. Wireless connections may be implemented using Wi-Fi, WiMAX and Bluetooth, infrared, cellular networks, satellite, or any other wireless connection method. Additionally, several networks may operate independently or communicate with each other to facilitate communication within network 971.

[0107] The functions and process steps described herein may be executed automatically or in whole or in part in response to user commands. Automatically executed activities (including steps) are performed in response to one or more executable instructions or device operations without requiring the user to directly initiate the activity.

[0108] The methods and systems described herein can be embodied, at least in part, in the form of computer-implemented processes and means for performing these processes. The disclosed methods can also be embodied, at least in part, in the form of a tangible, non-transitory machine-readable storage medium encoded with computer program code. This medium may include, for example, RAM, ROM, CD-ROM, DVD-ROM, BD-ROM, hard disk drive, flash memory, or any other non-transitory machine-readable storage medium, wherein, when the computer program code is loaded into and executed by the computer, the computer becomes a means for performing the method. The method can also be embodied, at least in part, in the form of a computer, in which the computer program code is loaded and / or executed, making the computer a dedicated computer for performing the method. When implemented on a general-purpose processor, the computer program code segments configure the processor to create specific logic circuits. Alternatively, the method can be embodied, at least in part, in a digital signal processor formed from application-specific integrated circuits for performing the method.

[0109] Therefore, according to one aspect of the present invention, a non-transitory machine-readable storage medium encoded with computer program software is disclosed, such that when a processor executes the computer program software, the processor performs a method for TA in a CBM PET scanner, wherein the method includes: (a) Receive list pattern data from a PET scan of an axial short phantom in CBM mode; (b) Generate CBM non-time-of-flight (TOF) projection data, scanner efficiency normalized array n, and attenuation factor a; (c) Reconstruct image f from CBM non-TOF projection data, correct for scanner efficiency by normalizing array n, and correct for attenuation by attenuation factor a; (d) Non-TOF and TOF scattering of motion blur and non-TOF and TOF projection of motion blur Perform modeling; (e) For each response line, the zeroth moment from the modeled TOF data and first moment Estimate the TOF centroid (COM) of the true coincident distribution collected in the stationary coordinate system; (f) For each response line, calculate the measured TOF COM; and (g) Determine the TOF time offset by taking the difference between the measured TOF COM and the modeled TOF COM. to .

[0110] According to some embodiments, a non-transitory machine-readable storage medium encoded with computer program software is disclosed, such that when a processor executes the computer program software, the processor performs a method for monitoring the performance of a PET scanner during CBM acquisition, wherein the PET scanner has a patient bed that moves an axially short cylindrical phantom via axial motion during CBM acquisition, the method comprising: (a) Time-of-flight (TOF) data were generated during CBM acquisition of the axial short phantom, taking into account the axial motion of the patient bed during CBM acquisition, as the first dataset; (b) A supplementary dataset was generated by integrating the TOF data on the axial motion of the patient bed during CBM acquisition; and (c) Reconstruct the activity and crystal efficiency normalization coefficients simultaneously from the supplementary dataset.

[0111] Applications of axial short phantoms The axially short phantom described herein for use in CBM TA procedures and CE estimation has a shorter axial length compared to the FOV of a PET scanner. The axially short phantom of this disclosure comprises a cylindrical shell with a conventional length, for example, approximately 28 cm. However, the central portion, shorter than the FOV of a PET scanner, is filled with activity. "Axially short" refers to the fact that the activity-filled portion of the phantom is shorter than the FOV of a PET scanner. Figure 7A An example of such an axial short phantom 60, placed on a standard bracket mounted on a CBM bed 260, is shown in accordance with this disclosure.

[0112] Figure 7B The CBM data reconstruction is shown in the activity, cross-sectional (left), and coronal (right) views.

[0113] Figure 7C The image shows CT-based attenuation maps, with cross-sectional (left) and coronal (right) views.

[0114] The remainder of the phantom shell is filled with air, creating a buffer between the activity wire and the metal support that connects the phantom to the CBM bed. The axially short phantom is designed so that the activity LOR does not pass through the CBM bed or the phantom support. The axially short phantom disclosed herein is approximately lighter than a standard, conventionally long phantom. This makes it easier to handle.

[0115] Experimental data used for confirmation A CBM TA procedure was performed using an axially short phantom and compared with a standard TA procedure using a conventional 28 cm long phantom. The CBM setup with the axially short phantom of this disclosure was tested on a Siemens Vision PET / CT scanner. There were 760 (cross-sections in one ring) x 80 (axial ring) detector crystals. List mode information consisted of a 13 ps TOF chamber. During the TA procedure, the time window was opened to 6.7 ns to avoid data truncation in the time direction.

[0116] Before acquiring CBM data from the axial short phantom used for comparison, a standard TA procedure was performed using a standard 28 cm long phantom. The list pattern scan consisted of 250 million truth counts. The result of this procedure was a TO array, which was downloaded by firmware for TOF bin index correction during the following list pattern data acquisition in a 4.7 ns time window.

[0117] CBM scanning was performed using an axially short phantom, eliminating the need to download estimated TOs (without TA) for acquiring list-pattern data. Scans were performed at a speed of 0.3 mm / s. At the start and end of the data acquisition scan, the axially short phantom was completely outside the scanner's FOV. This resulted in approximately 352 million true counts in the "static" data. Once TOs were estimated, they were applied during list-pattern data processing by modifying the TOF bin indexes. The modified list-pattern file was used for CE estimation.

[0118] For comparative purposes, the CBM data are also treated as static scans. Bed motion is ignored to produce long virtual cylinder scans (from the perspective of activity and decay). The dataset underwent standard procedures for TO and CE estimation.

[0119] According to the National Electrical Manufacturers Association (NEMA) standard, the line source phantom is placed at the center of the scanner to evaluate time resolution. List pattern data is acquired without downloading the Time of Execution (TO). During list pattern data processing, the TO is applied by changing the TOF bin index. In this way, the same dataset is used to estimate the performance of various time-of-flight (TA) procedures.

[0120] Figures 8A to 8D The modeling sine plot of the "static" scan data is shown, which is the motion-blurred attenuated projection of the axial short cylindrical phantom. Figure 8A and 8B The activity sine diagrams for different polar angle segments are shown. Figure 8A It is a direct plane sine curve. Figure 8B It is the most tilted sine curve. Figure 8C and Figure 8D The attenuation factor used in modeling the projected data is shown. Figure 8C It is the direct plane attenuation factor, coronal view. Figure 8D The most sloping attenuation factor is the coronal plane. It's worth mentioning that the direct and sloping sine plots have different structures due to the significant difference in attenuation. For sloping data, the activity LOR passes through not only the surface of the cylindrical phantom but also its bottom, which is the case for the direct plane LOR. This example illustrates that the assumption of a uniform virtual cylinder constructed by moving a short cylinder might be inappropriate. Because the phantom is outside the FOV at the beginning and end of acquisition, the polar segment of the projected data is uniform in the axial direction.

[0121] Figures 9A to 9C The detector crystal estimations performed using two different methods and their differences are shown. Figure 9A The TO estimate performed using the standard TA method is shown. Figure 9B The TO estimation performed using the CBM TA method is shown. Figure 9C The differences are shown. The differences are small and are mainly attributed to estimation noise. TO values ​​are in TOF bins (per 13 ps).

[0122] After applying standard TA TO, the line source TOF resolution is 205.32 ps. CBM TO results in a TOF resolution of 205.61 ps for the same dataset.

[0123] Figures 10A to 10E The CE estimates are shown for both standard and CBM settings. Small differences are observed as block edge patterns when the axially short cylindrical phantom is correctly modeled. For the virtual cylindrical phantom, the differences are non-uniform across the axial direction. The unique LOR, which is only an approximation, contributes nothing to the edges of the axial FOV where the direct plane has appropriate attenuation. This imbalance introduces inhomogeneity into the CE estimates. Figure 10A The CE estimation performed using standard methods is shown. Figure 10B The CE estimation using the CBM axial short phantom TO is shown. Figure 10C It shows Figure 10A and Figure 10B The differences between the CE estimates shown. Figure 10D The CE estimation using the virtual cylinder TO is shown. Figure 10E It shows Figure 10A and Figure 10D The differences between the CE estimates shown.

[0124] Figure 10F The axial curve of CE averaged on a detector module is shown. Line (1) represents... Figure 10A The CE estimate shown is represented by line (2). Figure 10B The CE estimate shown is represented by line (3). Figure 10D The CE estimate is shown.

[0125] Figure 11A and Figure 11B The propagation of CE estimation defects in cylinder image reconstruction was examined. Figure 11A The image shows the normalized difference between the uniform cylinder reconstruction derived using the CBM axial short cylinder phantom and the standard normalization. Figure 11B The image shows the normalization difference between a uniform cylinder reconstruction derived using a virtual cylinder and a standard normalized one. Figure 11A and Figure 11B In the diagram, the top view is the axial view, and the bottom view is the coronal view.

[0126] Standard QC acquisition with standard QC normalized image reconstruction was used as the gold standard. The CE components were then replaced in the normalized array, and the data was reconstructed again. Using the CE components obtained from the CBM axial short cylindrical phantom, image differences showed small circular artifacts. These are attributed to observed block edge patterns. Using CE component normalization derived from the virtual cylindrical phantom, image differences showed approximately 5% axial inhomogeneity, which can be attributed to the axial inhomogeneity of this CE component.

[0127] When PET scanners become longer in an attempt to gain sensitivity, phantoms with an axial length shorter than the scanner's axial FOV are a more desirable choice than manufacturing longer phantoms for system setup. Static scanning can be used to cover only a portion of the available LORs. Since covering all possible LORs is beneficial, multiple or CBM scans can be utilized. The presented results confirm that the CBM TA procedure using short, uniform cylindrical phantoms produces virtually the same TO correction compared to standard TA.

[0128] By employing CBM reconstruction, no assumptions are made about virtual (excluding bed motion) projection data. This allows for the use of more complex models with additional fine structures, which can also cover other purposes such as spatial alignment of PET and CT scanners.

[0129] Additional uses of axial short phantoms An axial short phantom in CBM mode of a PET scanner can be used to examine the normalization coefficient of the PET detector. By performing a PET scan while the axial short phantom moves through the patient tunnel of the PET gantry, the uniformity of the PET detector along the PET scanner's field of view (FOV) can be measured and used to determine the detector's normalization coefficient.

[0130] Furthermore, since the phantom has uniform and known photon output, the axial short phantom 60 in the CBM setup can be used to check the alignment of the CBM bed 260 with the PET detector by scanning the axial short phantom 60 in CBM mode. The method for checking the alignment of the axial movement of the CBM bed includes acquiring list pattern data from a PET scan of the axial short phantom in CBM mode; calculating the TOF centroid of the axial short phantom from the list pattern data; and comparing the calculated TOF centroid of the axial short phantom with the central axis of the PET scanner stage.

[0131] Furthermore, due to the lighter weight of the axially short phantom, it can be cantilevered and mounted on the CBM machine 260, such as... Figure 7A As shown. In this cantilever arrangement, the axial short phantom 60 is positioned away from and in front of the CBM bed 260 extension. The axial alignment of the CBM bed 260 can be checked by measuring and tracking its TOF COM as the axial short phantom moves through the PET bench.

[0132] The cantilever arrangement of the axial short phantom 60 allows for minimization of any attenuation effects on the CBM bed and phantom holding hardware 280.

[0133] While the subject matter has been described with reference to exemplary embodiments, it is not limited thereto. Rather, the appended claims should be interpreted broadly to include other variations and embodiments that may be made by those skilled in the art.

Claims

1. A method for time alignment in a continuous bed motion CBM positron emission tomography (PET) scanner (100), the method using an axially short phantom (60) to cover the field of view of a long axial PET scanner, wherein the axial length of the phantom (60) is shorter than the field of view of the PET scanner (100), and wherein the phantom (60) is cantilevered on a patient bed (260) such that the phantom (60) extends axially in a manner that is in front of the patient bed (260) away from the patient bed, the method comprising: (a) Receive list pattern data from a PET scan of an axial short phantom (60) in CBM mode; (b) Based on the list pattern data, generate CBM non-TOF projection data, scanner efficiency normalized array n, and attenuation factor a, where TOF represents time of flight; (c) Reconstruct image f from CBM non-TOF projection data, correct for scanner efficiency by normalizing array n, and correct for attenuation by attenuation factor a; (d) Based on the image f, perform non-TOF and TOF scattering S and non-TOF and TOF projection of motion blur. Modeling; (e) For each response line, from the modeled TOF projection zeroth moment and first moment Estimate the true coincidence distribution of the TOF COM sampled in the stationary coordinate system, the zeroth moment and first moment Depends on the non-TOF and TOF scattering S and the non-TOF and TOF projection COM represents the centroid; (f) For each response line, calculate the measured TOF COM; and (g) Determine the TOF time offset by taking the difference between the measured TOF COM and the modeled TOF COM. to The TOF time offset to It was used to calibrate the PET scanner (100).

2. The method according to claim 1, wherein, Step (e) is performed using the following expression: ； in , ; as well as ； Where k is the voxel index; Where (ρθζ) is the spatial projection index, representing the radial ρ, azimuth θ and axial ζ projections, which include the polar segment; Where T is the TOF warehouse index; in It is estimated TOF COM data. It is the TOF data used for modeling. yes The first moment, yes The zeroth moment of , and C is the geometric projection system matrix; as well as B is a fuzzy operator defined as follows: ， in d It is the dead time correction factor; δ It is the Kronecker notation; z is the axial plane coordinate of the scanner data, including the polar angle; Z ( ζ , t ) indicates that the coordinates A function mapped to scanner system coordinates, where t It refers to the data collection time; It is the decay correction constant for a specific isotope; and t acq This is the duration of CBM data collection.

3. The method of claim 2, wherein step (f) is performed using the following expression: ,in It is the measured TOF COM, and And among them It is real-time TOF data and It is random data.

4. A method for checking the normalization coefficient of a positron emission tomography (PET) detector in a continuous bed motion CBM PET scanner (100), comprising: A PET scan is performed while an axially short phantom (60) moves through the patient tunnel of a PET scanner (100), wherein the axial length of the phantom (60) is shorter than the field of view of the PET scanner (100), and the phantom (60) is mounted on a patient bed (260) in a cantilever arrangement such that the phantom (60) extends axially away from the patient bed in front of the patient bed (260); and The normalization coefficients of the detector are determined using measurement data and a supplementary dataset generated from the projection reconstructed by the CBM, including: (a) Time-of-flight (TOF) data of positron annihilation activity in an axially mounted short phantom on a patient bed, generated during CBM acquisition, taking into account the axial motion of the patient bed during CBM acquisition, are used as the first dataset; (b) A supplementary dataset is generated by integrating the TOF data over the axial motion of the patient bed during CBM acquisition, wherein the supplementary dataset is modeled by the following equation: ， in Calculated using the motion fuzz operator B by the following formula: ;as well as (c) Reconstruct the normalized coefficients of positron annihilation activity and crystal efficiency simultaneously from the supplementary dataset. in: It is CBM data; It is the average of the CBM dataset; It is the projection of the model; is the average random data; n is the normalized array; (ρ θ z) is the spatial projection index, representing the radial projection ρ, azimuth θ, and axis-plane coordinate z including the polar angle; (ρ θζ) is the spatial projection index, representing the radial ρ, azimuth θ, and axial ζ projections, including the polar angle segment; T is the TOF bin index; B is the fuzzy operator; S is the scattering estimate; a is the attenuation factor; C is the geometric projection system matrix; f is the emission object; k is the voxel index; It is the crystal efficiency; d It is the dead time correction factor; δ It is the Kronecker symbol; Z ( ζ , t ) indicates that the coordinates A function mapped to the scanner system coordinates, where t is the acquisition time; It is the decay correction constant for a specific isotope; and t acq This is the duration of CBM data collection.

5. A method for monitoring the performance of a positron emission tomography (PET) scanner during continuous bed motion CBM acquisition, the method using an axially short phantom (60) to cover the field of view of a long axial PET scanner, wherein the PET scanner (100) has a patient bed (260) through which a patient is axially moved during CBM acquisition, and the axial length of the phantom (60) is shorter than the field of view of the PET scanner (100), and the phantom (60) is cantilevered on the patient bed (260) such that the phantom (60) extends axially in a manner that is in front of the patient bed (260) away from the patient bed, the method comprising: (a) Time-of-flight (TOF) data of positron annihilation activity in an axially mounted short phantom on a patient bed, generated during CBM acquisition, taking into account the axial motion of the patient bed during CBM acquisition, are used as the first dataset; (b) A supplementary dataset is generated by integrating the TOF data over the axial motion of the patient bed during CBM acquisition, wherein the supplementary dataset is modeled by the following equation: ， in Calculated using the motion fuzz operator B by the following formula: ;as well as (c) Reconstruct the normalized coefficients of positron annihilation activity and crystal efficiency simultaneously from the supplementary dataset. in: It is CBM data; It is the average of the CBM dataset; It is the projection of the model; is the average random data; n is the normalized array; (ρ θ z) is the spatial projection index, representing the radial projection ρ, azimuth θ, and axis-plane coordinate z including the polar angle; (ρ θζ) is the spatial projection index, representing the radial ρ, azimuth θ, and axial ζ projections, including the polar angle segment; T is the TOF bin index; B is the fuzzy operator; S is the scattering estimate; a is the attenuation factor; C is the geometric projection system matrix; f is the emission object; k is the voxel index; It is the crystal efficiency; d It is the dead time correction factor; δ It is the Kronecker symbol; Z ( ζ , t ) indicates that the coordinates A function mapped to the scanner system coordinates, where t is the acquisition time; It is the decay correction constant for a specific isotope; and t acq This is the duration of CBM data collection.

6. The method of claim 5, wherein step (a) includes having a spatial projection index. The CBM data y is used to model the TOF warehouse index T by combining the following items: the projection of the real event model from the emission object f. Defined by the voxel index k, with a corrected for attenuation by the normalized array n for scanner efficiency; and a scattering estimate S, corrected for scanner efficiency; and average random data. : ； Where coordinates It is CBM data normalization.

7. The method of claim 6, wherein the CBM data normalization factor It is calculated based on the scanner normalization factor Z, taking into account axial patient bed movement during CBM acquisition, according to the following equation: ； in It is the decay correction constant for a specific isotope. It is a Kronecker notation, d is the dead-time correction factor, and t acq This is the CBM acquisition time.

8. The method of claim 7, wherein the scanner normalization factor Z is adaptable to the mixing and recombination of response lines, and the two detector scintillation crystals i and i' are connected to the projection chamber according to the following equation: ； Where g is the geometric component of the normalized array; and in This is due to the contribution factor of the response line into the projected data caused by the hybrid and cross-data compression techniques, and Defined by the following formula: , Where j is the spatial projection index, which consists of the radial projection ρ, the azimuth angle θ, and the axis-plane coordinate z including the polar angle.

9. The method of claim 5, wherein step (c) comprises: Regarding f, maximize the following objective function For use in life restoration; Where f is the activity graph; This involves applying the ordered subset expectation maximization algorithm using one iteration and 21 subsets; and about Maximize the following objective function , Where j is the spatial projection index, which consists of the radial projection ρ, the azimuth angle θ, and the axis-plane coordinate z including the polar angle.

10. The method of claim 5, further comprising updating the crystal efficiency normalization coefficient by the following update. ， Where m is the crystal renewal efficiency The number of iterations; Where g is the geometric component of the normalized array; in This is due to the contribution factor of the response line into the projected data caused by the hybrid and cross-data compression techniques, and Defined by the following formula: ； in and Indicates the index T in the TOF repository. and The sum; and Where j is the spatial projection index, which consists of the radial projection ρ, the azimuth angle θ, and the axial plane coordinate z including the polar angle; and i and i' represent the two detector scintillation crystals.

11. A positron emission tomography (PET) calibration system, comprising: A PET scanner (100) having an axially short phantom (60) mounted on a patient bed (260) that moves axially via the PET scanner during CBM acquisition, wherein the axial length of the phantom (60) is shorter than the field of view of the PET scanner (100), and wherein the phantom (60) is mounted on the patient bed (260) in a cantilever arrangement such that the phantom (60) extends axially away from the patient bed in front of the patient bed (260); and A scintillation crystal efficiency calibration system that performs the method according to any one of claims 5-10.

12. A non-transitory machine-readable storage medium encoded with computer program software, such that when a processor executes the computer program software, the processor performs the method according to any one of claims 1-10.

Images