CN113116371A - PET device and calibration method - Google Patents

Abstract

Embodiments provide a PET device and a calibration method capable of performing calibration with good accuracy. The detector detects pairs of simultaneous events and is comprised of a plurality of processing units. The circuit corrects relative timing offsets within the group of the processing units based on data acquired via a radiation source, corrects the timing offsets between the groups of the processing units using intrinsic radiation, and performs calibration of the detector based on the corrected relative timing offsets within the group of the processing units and the corrected timing offsets between the groups of the processing units.

Description

PET device and calibration method

Reference to related applications

The present application enjoys the benefit of priority of U.S. patent application No. 62/955270, filed on 30/12/2019, U.S. patent application No. 16/907972, filed on 22/6/2020, and japanese patent application No. 2020-211197, filed on 21/12/2020, the entire contents of which are incorporated herein by reference.

Technical Field

Embodiments disclosed in the specification and drawings relate to a PET apparatus and a calibration method.

Background

As a method OF performing timing calibration OF a detector in a TOF (Time OF Flight) PET (Positron Emission Tomography) apparatus, there is a method OF using an external radiation source such as a phantom (phantoms). However, methods of using external sources such as phantoms to perform timed calibration of detectors have the disadvantage that the phantoms are large, difficult to handle, or require complex mechanical motion.

As another method of timing calibration of a detector in a TOF PET apparatus, a method using intrinsic radiation (intrinsic radiation) of the detector is also considered. However, methods for timing calibration of detectors using the inherent radiation of the detector generally require a very long time.

Disclosure of Invention

One of the technical problems to be solved by the embodiments disclosed in the present specification and the accompanying drawings is to perform calibration with good accuracy. However, the technical problems to be solved by the embodiments disclosed in the present specification and the drawings are not limited to the above technical problems. Technical problems corresponding to the respective effects of the respective configurations shown in the embodiments described below can be also positioned as other technical problems.

The PET device according to the embodiment includes a detector and a circuit. The detector detects a pair of coincident events generated by annihilation (annihilation) of positrons emitted from a radiation source disposed within a field of view (FOV) of the detector, and is configured by a plurality of processing units. The circuit corrects relative timing offsets within the group of the processing units based on data acquired via the radiation source, corrects the timing offsets between the groups of the processing units using the intrinsic radiation, and determines a total timing offset based on the corrected relative timing offsets within the group of the processing units and the corrected timing offsets between the groups of the processing units, thereby performing the calibration of the detector.

ADVANTAGEOUS EFFECTS OF INVENTION

According to the PET apparatus of the embodiment, calibration with high accuracy can be performed.

Drawings

Fig. 1 is a schematic diagram of a PET scanner according to an embodiment.

Fig. 2A is a schematic diagram illustrating an exemplary position of an external radiation source relative to opposing detectors.

Fig. 2B is a diagram representing a TOF difference histogram between the 2 detectors of fig. 2A.

FIG. 3 is a diagrammatic view illustrating another exemplary position of an external radiation source pair relative to opposing detectors.

FIG. 4 is a diagrammatic view illustrating another exemplary position of an external radiation source relative to opposing detectors.

FIG. 5 is a diagrammatic view illustrating another exemplary position of an external radiation source relative to opposing detectors.

Fig. 6A is a schematic diagram illustrating various embodiments of an external positron emission source within a detector ring.

Fig. 6B is a schematic diagram illustrating various embodiments of external positron emission sources within a detector ring.

Fig. 6C is a schematic diagram illustrating various embodiments of external positron emission sources within a detector ring.

Fig. 6D is a schematic diagram illustrating various embodiments of external positron emission sources within a detector ring.

FIG. 7A is a diagrammatic view illustrating various embodiments of an internal radiation source within a detector ring.

FIG. 7B is a diagrammatic view illustrating various embodiments of an internal radiation source within a detector ring.

Fig. 8A is a diagram showing an internal radiation timing distribution before applying correlation offset correction obtained from an external radiation source.

Fig. 8B is a diagram showing a distribution relating to the same data after the walk correction (including the non-linear walk correction) is applied.

Fig. 9 is a diagrammatic view of an exemplary configuration showing various stages of timing calibration.

FIG. 10 is a diagrammatical illustration of another exemplary configuration showing various stages of timing calibration.

Fig. 11 is a diagram showing an ascending discriminator used in a TOF PET system.

Fig. 12A is a graph showing a TOF difference histogram between DU14 and DU33 from Lutetium background radiation data (Lutetium background radiation data).

Fig. 12B is a graph representing a TOF difference histogram between DU14 and DU33 from lutetium background radiation data.

Fig. 13 is a graph representing a TOF difference histogram between DU14 and DU33 from lutetium background radiation data.

Fig. 14 is a graph showing TOF difference histograms between DU14 and DU33 from the Ge line source and the background radiation data.

Fig. 15A is a diagram showing timing offsets between DU pairs before short-time timing calibration.

Fig. 15B is a diagram showing timing offsets between DU pairs after short-time timing calibration.

Fig. 16 is a diagram showing the difference in count that can be used between the case where the count is limited by applying an energy window in order to avoid the influence of walk (walk) and the case where all counts are accepted by performing walk correction.

Detailed Description

Hereinafter, embodiments of the PET apparatus and the calibration method will be described in detail with reference to the drawings.

In PET, an image of a living body is created. PET scanners produce images representing various biological processes and functions. Typically, in a PET scan, a patient is initially administered a radioactive substance called a PET isotope. The delivered PET isotope may function as a tracer if it is related to a standard (japanese patent) physiological process in the body of the patient. As a typical positron emission PET isotope, mention may be made of¹¹C、¹³N、¹⁵O and¹⁸F. if a positron (and a neutron) is emitted from an unstable nucleus in the body, it is combined with an electron in the adjacent tissue and annihilated. By annihilation events, a pair of gamma photons (gamma photons) are generated that are emitted in opposite directions. The gamma photons are detected by the detector ring as shown in fig. 1. The detector ring 300 may be composed of a plurality of detectors (e.g., 101) each including a scintillator block and a photon sensor. For example, the detector 101 is constituted by a scintillator block 102 and a photon sensor 103.

One mode of PET detection is TOF PET, which measures the arrival times of a pair of coincident photons. In TOF PET, when a radiation event (e.g., a gamma photon) is detected, a scintillator block at the detection site immediately records the time of the detected radiation event. By obtaining the arrival time, more weight is given to a place where the probability of the radiation position of each event is higher, thereby reducing the statistical uncertainty of the reconstructed image.

In PET, each detector assigns an energy and a time stamp to each detected gamma ray. An energy window is applied to select energies in a range around 511keV, while a coincidence timing window is applied to determine coincident pairs of gamma rays. A simultaneous counter Line (LOR) connecting 2 detector elements for detecting gamma rays is defined by each simultaneous pair. To generate an image from the detected LORs, a reconstruction method is applied.

In TOF PET, using the difference in the time stamps of the respective gamma rays of the coincident pairs, a higher weight is given to a place where the probability of the annihilation position of each event is higher, thereby reducing the statistical uncertainty of the reconstructed image. To provide significant improvement in reconstructed images, the time difference (time difference) of measurement for each LOR record is typically in the range of hundreds of picoseconds and must be very accurate. Calibration requires that the accuracy of the measured time difference be sufficiently ensured, because of unavoidable manufacturing tolerances, such as cable lengths or timing response differences of different photon sensors within the detector.

As one method, it is considered to individually calibrate the time difference of each LOR in the system. However, PET scanners typically have up to tens of thousands of detector elements, with LORs in the order of hundreds of millions, and this approach is therefore not practical. The general approach of the prior art is to perform offset correction on each detector element. Before determining whether or not the detected gamma rays are part of the simultaneous pairs for each detector element, an offset correction value is given to the measurement time stamp of that element. Offset correction values with respect to 2 gamma rays of a simultaneous pair are also applied to calculate the TOF difference of the pair used in the reconstruction. Typically, the timing offset value is a "signed" value, meaning that there are positive or negative conditions.

In a TOF PET system for measuring a timing difference (timing difference), when N detector elements are present in the system, (N-1) timing offset values need to be determined in order to enable accurate measurement of the timing difference. For example, a timing offset value for a certain element may be arbitrarily set to zero, and other (N-1) offset values may be determined based on the element arbitrarily set to zero. Alternatively, the N offset values may be calculated under an additional condition (again, meaning that there are only (N-1) independent values) in which the average offset value for all of the N detector elements is zero. In this way, when the same additional timing offset value is given to all the measurement values, any measurement timing difference does not change.

The following are some non-limiting examples of the above. These examples are merely illustrative and do not necessarily represent preferred embodiments of a clinical PET system.

As shown in fig. 2A, an example of an annihilation radiation source 50 disposed between 2 detector elements D10 and D30 is examined. By coincident events, 1 LOR60 is generated between detector elements. In this example, the annihilation radiation source 50 is positioned equidistant from D30 from D10 (other than equidistant options are equally effective). The time measured by D10 is denoted by t_D10The time measured by D30 is denoted by t_D30And (4) showing. From the noise or uncertainty in the time measurement, a time difference (t) is made_D10-t_D30) Is sometimes plotted as a straight line as shown in fig. 2BA block diagram.

Prior to calibration, the histogram 70 is determined to have t_measuredWhich is sometimes calculated as a pure average or determined, for example, by least squares fitting to a gaussian function of the measured histogram. Since the source is equidistant from the 2 detectors, the time-of-flight differences relative to the 2 detectors are equal, with the expected time difference being zero.

By measuring the time difference distribution, a single equation, (t) can be written_D10+t_OffsetD10)-(t_D30+t_OffsetD30)＝t_measured+t_OffsetD10-t_OffsetD30＝0。

In the formula, t_OffsetD10And t_OffsetD30Offset correction values for D10 and D30, respectively. Since there are only 1 equation and 2 unknowns, t can be determined_OffsetD10And t_OffsetD30Other conditions need to be applied.

As 1 example, one may consider selecting t_OffsetD10As a result, t is 0_OffsetD30＝t_measuredThe case (1). As another example, (t) is selected_OffsetD10+t_OffsetD30) 0/2, as a result, t_OffsetD10＝-t_measured/2, and t_OffsetD30＝t_measuredThe case of/2.

By selecting any one of these 2 (or other possible selection of conditions), the corrected timing histogram 80 centered on the expected value (zero) is obtained as a result. In this simple example, 2 detectors (i.e., N-2) are used, 1 of which can determine 1 (i.e., N-1) independent offset values, and the system can be calibrated to accurately measure the timing difference (i.e., TOF).

When the number of detectors is increased to 4 and the annihilation radiation sources are placed between the detectors at equal distances from them, 2 LORs are generated, and as a result, 2 equations relating to the measured average time difference for each LOR respectively are obtained. For a detector with N-4, 3 (N-1-3) independent offset values need to be determined for full timing calibration. It is clearly impossible to determine 3 unknowns from 2 equations. This means that it is a substantial problem to perform timing calibration using a single limited range radiation source at a single fixed position.

The 1 method for solving this problem is shown in fig. 3. The second radiation source 55 is placed equidistant between the detectors D30 and D40 (again, the equidistant option is used to simplify the illustration, not necessarily the option). With this second radiation source, additional LORs 65 are generated, so a third equation can be written relating timing offset values relative to each other. By obtaining 3 equations with 3 unknowns, it is possible to solve for 3 (N-1-3) independent offset values needed to calibrate the timing response of the system over all. The respective pairs of detector elements can be paired by a series of LORs (with 4 detector elements, there are 6 possible combinations of detector elements), i.e.,

d10 was directly paired with D30 via LOR60,

d20 was directly paired with D40 via LOR62,

d30 was directly paired with D40 via LOR65,

d10 was indirectly paired with D40 via LOR60 and LOR65 (via D30),

d10 was paired indirectly with D20 via LOR60, LOR65 and LOR62 (via D30 and D40),

d20 is indirectly paired with D30 via LOR62 and LOR65 (via D40), and therefore, all timing differences required for the system can be determined.

This example means that, when there are 3 or more detectors, a certain degree of "one-to-many" pair of elements is necessary for measurement. In this case, the additional radiation source brings a pair of detectors D30 and D40.

Fig. 4 and 5 show 2 examples of one-to-many pairs in a group (group) of detectors, and it is possible to determine a necessary independent offset value in a group (group) associated with an additional pair. For example, in fig. 5, LOR67, which concatenates the first group of detectors, can calculate 5 independent offset correction values. Similarly, LOR69 connecting the second set (set) of detectors can calculate 5 additional independent offset values, and the total number of offset values determined is 10. This total of 10 is missing 1 relative to the 11 independent offset values required. This lack occurs because there are no cross-pairs between 2 sets (sets) based on the LOR of a single limited-range radiation source 50. This example means that a group (group) or a one-to-many pair within a group (set) is able to determine the relative offset values within the group, but a crossing pair of groups (sets) is also necessary to calibrate the timing response of the system in its entirety. The requirements with respect to one-to-many pairs and to crossing pairs between sets of detector elements are dealt with in the prior art by several methods as described later.

In a full timing calibration, one-to-many pairs of all detector elements that are calibrated are required. Previous calibration methods generally used external and/or internal radiation sources.

Several methods of using external radiation sources are shown in figures 6A to 6D. These methods all implement the required one-to-many pairs, including crossing pairs across sets of detector elements. For example, the detector elements within a set (set) are sometimes individual scintillator elements within a scintillator array of a single detector module. In this case, the cross-pairing means that the described external radiation source method provides pairs across the inter-module gap, i.e., LORs based on several scintillator elements paired with scintillators in 2 or more detector modules on the opposite side of the detector ring.

Hereinafter, a phantom refers to a specially designed object placed in the field of view of the detector of a scanner for scanner calibration or scanner performance evaluation. In PET, the phantom typically contains a positron-emitting source (e.g., Ge-68, F-18, or Na-22) and, in many cases, a surrounding material, the emitted positrons are reliably converted to back-to-back (back-to-back) 511keV photons in a short period of time by annihilation with electrons in the surrounding material. Phantoms are sometimes as simple as a simple stent or mount (mount) for a radiation source. The phantom may also include a material intended to scatter or partially absorb the emitted radiation. For example, a cylindrical mold is often used for PET. The fillable cylindrical mold body is sometimes constructed from a cylinder of acrylic with a central void and a closable port. In use, the central void is filled with a radioactive liquid such as Fluorodeoxyglucose (FDG) (labelled with F-18) mixed with water. Alternatively, a cylindrical phantom enclosing the radiation source is sometimes used. In this case, the plastic cylindrical shell is sometimes loaded with Ge-68 or the like isotopes and then filled with a cured epoxy that cures. The cylindrical mold body radiates radiation, and the material of the cylinder also scatters and attenuates the radiation. The phantom includes movable portions for calibration purposes or to simulate the movement of organs such as the beating or breathing movements of the heart.

Fig. 6A shows a method of using a large cylinder mold body 51. For example, the phantom is sometimes about 20cm in diameter, the same length as the field of view of the scanner in the direction of the body axis, and is the case with a cylinder filled with an epoxy containing Ge-68. As shown by the representative LOR61, each LOR via phantom pairs each scintillator element within the scintillator block 102 with a plurality of other scintillator elements including scintillator elements within several different detector modules 101. The method shown in fig. 6A has disadvantages in that it is difficult to handle since there is a case where the mold body itself has a weight of 20kg, and in addition, strict shielding (e.g., about 150kg of lead) is required in order to protect staff and patients when the mold body is not used.

Fig. 6B illustrates a method of using a movable mold body 52. In this case, the phantom may be composed of a rod-shaped radiation source (for example, Ge-68 in a steel sleeve) for annihilating radiation and a device for moving on an annular orbit in the field of view of the detector of the scanner. As the source rotates, the representative LOR61 pairs each scintillator element within the scintillator block 102 with a plurality of other scintillator elements, including scintillator elements within several different detector modules 101. The complexity cost and maintenance of the devices that bring about the movement of the radioactive source are drawbacks of this approach.

Fig. 6C shows a method of using a phantom composed of a cylindrical annihilation target 53 and a positron emission source (for example, Ge-68, not shown) which is not shielded from other sources, and as a result, positrons are separated from the emission source and annihilated with electrons in the annihilation target 53 to generate back-to-back 511keV annihilation radiation. For example, the annihilation target may be a plastic cylindrical shell having a diameter of about 20cm and a length equal to the field of view of the scanner in the direction of the body axis. As with the method described above, each scintillator element within the scintillator block 102 is paired with a plurality of other scintillator elements, including scintillator elements within several different detector modules 101, via a LOR of phantom, indicated at 61. Larger size annihilation targets are sometimes difficult to handle and therefore are a disadvantage to this approach. Storage of larger sizes when not in use is also inconvenient.

Finally, fig. 6D shows a method in which one or more pairs of radiation sources for annihilation radiation are surrounded by a large bulk of the scattering medium. In this case, the phantom 54 sometimes contains a Ge-68 rod radiation source surrounded by a steel cylinder from an inner diameter to about 10 cm. There are several disadvantages to this approach. First, the phantom is quite heavy and sometimes difficult to handle. Next, because of the low efficiency in the attenuation and scattering of 511keV gamma rays, only a very small number of gamma rays emitted from the radiation source can be actually used to form a one-to-many pair, and thus the data collection time is long and the convergence of the iterative method of estimating the offset sometimes takes time. Furthermore, the accuracy, particularly the time axis resolution (200 ps) that can be achieved by the latest system, is reduced due to the uncertainty of the scattering location.

A conventional method using an internal radiation source is shown in fig. 7A to 7B.

FIG. 7A illustrates a method of pairing elements within the scintillator block 102 with elements around the detector ring using intrinsic radiation, as shown by a representative LOR 61. In general, time is required for data collection in order to achieve sufficient accuracy. In addition, since the intrinsic radiation from Lu-176 forms a very broad spectrum, the energy range of the identified events is limited to 2 fairly narrow windows (1 about 511keV, the other 1 about 307keV), when 1 of the stronger radiation of Lu-176 is present. As a result of the application of these energy windows, the available counts are extremely reduced.

The results are shown in fig. 16. The data shown in this figure are data taken from a PET system using a scintillator crystal (crystal) containing Lu-176. The total Lu-176 simultaneous energy spectrum is represented by 900. In the totipotent 900, the counts are 78, 416, 224. The simultaneous energy spectrum after application of the 2 energy windows is denoted by 950. The width of the energy spectrum 950 is significantly reduced because of the lower simultaneous probability that 1 event of a simultaneous pair is a 511keV window (in this case 435 to 625keV) and the other 1 event is a 307keV window (in this case 250 to 350 keV). In the energy window spectrum 950, the counts are 10, 241, 786. Thus, the application of 2 energy windows works to reduce the walking results of measurement timing drift, but also reduces the available counts to about 13% of the totipotent 900.

Fig. 7B shows another method of using intrinsic radiation. In this method, LORs linking adjacent scintillator blocks 102 as shown by a representative LOR61 are used. In addition to the difficulty of long data collection times, there is another significant disadvantage in limiting the LOR to contiguous modules. All scintillator elements need to be connected to the scintillator elements of the neighboring block by LOR, but the transmission of the emitted intrinsic radiation is limited (mainly 202keV and 307keV gamma rays, which are rather lower energy than the 511keV gamma rays detected by PET). Thus, the method is limited to the calibration of blocks smaller than approximately 20mm width.

In summary, the disadvantages of the prior art TOF PET technology are roughly described in 2 ways. With respect to methods using external sources, the phantom is large, difficult to handle, or requires complex mechanical movements. On the other hand, methods using intrinsic radiation are generally very time consuming.

The embodiments presented in this specification are significant new approaches, utilizing a single stationary "limited range" external radiation source in combination with intrinsic radiation. The calibration method divides the timing calibration into 2 steps. In a first step, the "relative timing offsets" within a group (group) of processing units are taken using a single stationary limited-range external radiation source. In the second step, the offsets between the groups (groups) of the processing units are acquired using the "intrinsic radiation". In this case, the total offset is the sum of the "relative timing offset" and the "processing unit offset".

The major disadvantages of the external radiation source approach in the prior TOF PET approach are eliminated by using a single stationary limited range external radiation source. Further, by dividing the processing into 2 steps, the total amount of data required for the unique irradiation processing can be significantly reduced, thereby eliminating the main disadvantage of the unique irradiation method in the conventional TOF PET.

The number of counts required for the intrinsic radiation step is significantly reduced for 2 reasons. First, by determining the relative offset in the external radiation source step, the offset thus determined in the intrinsic radiation step is simply the "processing unit offset". This means that all counts from all scintillator crystals within the processing unit can be collected, thereby reducing the collection time. As an example, in the case of a scintillator array of 1 processing unit with a crystal of 10 × 10 ═ 100, the number of counts required to achieve the desired accuracy at the processing unit level is reduced by a factor of 100 compared to the number of counts to achieve the same accuracy at the crystal level in the external radiation source step. Further, by using the intrinsic radiation timing data based on the result of the external radiation source step by the pre-correction, the width of the distribution of the intrinsic radiation data is greatly reduced. This effect becomes further large in the case where the relative displacement correction obtained in the external transmission source step includes the walk correction and particularly the nonlinear walk correction.

The offset values described above are associated with variations in the delay of the various components of the detector system. The 1 generation source of such a variation may depend on the energy in the timing discriminator that generates the timing signal. Fig. 11 shows the use of a large number of rising discriminators in a TOF PET system. The time t is given when the signal level exceeds a threshold value. Fig. 11 shows an example of signals simultaneously generated by 3 different energy gamma rays interacting with the detector elements. 3 different signals corresponding to energy E₁、E₂And E₃(E₁＞E₂＞E₃). The time (t respectively) at which the threshold value is exceeded although 3 gamma rays arrive at the same time₁、t₂And t₃) Different. This phenomenon is often referred to as timed walking.

The correction of the motion before the "natural radiation" step enables the use of all counts obtained, compared to the case of a limited analysis of events in a narrow energy window as shown in fig. 16 (which is currently performed in the prior art). This further reduces the required collection time.

Fig. 8A shows the intrinsic radiation timing distribution before applying the relative shift correction obtained from the external emission source, and fig. 8B shows the distribution relating to the same data after applying the timing correction (including the non-linear walk correction) from the first step. In this case, the width of the timing distribution is reduced by 2 times or more. To achieve the same statistical uncertainty by processing unit offset, the effect of this reduction in distribution width is roughly equivalent to increasing the number of counts by more than 4 times. Therefore, the data is corrected in advance by the second step before the process is divided into 2 steps and the calculation is performed, whereby the amount of data (or the collection time) required in the inherent irradiation step is reduced to approximately 100 × 4 to 400 in the presented example (compared with the case where the inherent irradiation of the crystal level is used, but the data is not corrected in advance). Thus, the collection time is changed from several hours when the intrinsic radiation is used for all calibration to about 1 minute in the method described in the present specification, and the collection is performed without using a large or movable external radiation source.

In timing offset calibration, all the processing units to be calibrated need to be paired by the simultaneous occurrence of events. Sometimes 1 processing unit is 1 element in any stage of the electronic device structure of fig. 9 and 10. The calibrated processing units are sometimes paired directly by coincident events. The calibrated processing units are sometimes also paired indirectly by coincident events. For example, processing unit 1 is paired with unit 2, unit 2 is paired with unit 3, and unit 1 is indirectly paired with unit 3.

To calibrate relative crystal shifts and walk-through within the processing unit, a small, lightweight limited-range external emission source (e.g., a small, finite diameter line source) can be used (without movement of the radiation source or multiple radiation sources).

Intrinsic radiation (even at low activity levels) can be used (after correction of relative crystal shifts and walk-through) for measuring process unit shifts. The pre-correction of relative crystal shift and walk narrows the initial timing distribution, and as a result, reduces the number of counts required to achieve a particular processing unit shift accuracy, making the collection time for the native radiation appropriate. Furthermore, for example, instead of a narrow range limited to energies around 511keV, all events from the intrinsic radiation can be used by the walk correction.

In this case, the total offset is the sum of the relative crystal offset and the processing unit offset.

Advantages include reducing the necessity for a large limited range radiation source, movement of the radiation source, or multiple radiation sources. Small radioactive source with limited range, simple treatment and shielding and low replacement cost. Furthermore, the processing unit offset can be periodically recalibrated using only the intrinsic radiation.

During an initial full timing calibration, the timing within the opposed pairs of processing units is shifted and the timing is moved, for example by centering the scanner about annihilation radiation, e.g., positron emitting sources. The radiation source needs to have a sufficient thickness so that each stage of the crystal is paired with 2 or more stages of the crystal in the opposed processing units. After correcting for relative timing offset and timing walk within the opposing pairs of processing elements, the timing offset between the processing elements is calibrated using intrinsic radiation within the crystal.

Here, the intrinsic radiation is radiation resulting from decay (decay) of a radioactive material (within a crystal, on the surface of a crystal, within a reflector material, etc.) that is part of the scintillator array. Typical intrinsic radiation is background radiation from naturally occurring isotopes of the scintillator material. Lu-176 from LYSO is an example of background radiation. Intrinsic radiation is sometimes intentionally added or doped to the scintillator material. As another example, Co-60 is a material that can be added to scintillator materials.

As for the requirements relating to the inherent radiation, there are: the decay process comprises at least 2 (approximately) simultaneous emissions (e.g. beta immediately after gamma) at times when a coincidence event occurs, or the decay process comprises 1 emission at times when a coincidence event occurs due to compton scattering within the detector caused by the emission, a half-life of more than 10 years and thus having a natural activity with an effect throughout the lifetime of the scanner, the activity relative to each 1cm³In the range of 100 to 1000Bq and thus the data collection time is practical and the occurrence of sporadic events is excessive, the energy of the radiation is several hundred keV to 1MeV, and Lu-176 and Co-60, etc. are cited as candidates.

During daily clinical use, timing offset corrections based on prior timing calibrations and timing walk corrections are applied prior to recalibration of timing offsets at the processing unit level. The timing offset for each processing unit is sometimes calculated together using the intrinsic radiation and the annihilation radiation, e.g., the central positron emission source, or the timing offset for each processing unit is sometimes calibrated using the intrinsic radiation during periods when the scanner is not in use. In some cases, 1 processing unit is 1 element in an arbitrary stage (stage) of the electronic device configuration of fig. 9 and 10.

Throughout this specification, reference to "1 embodiment" and "one embodiment" means that a specific part, structure, material, or feature described in connection with the embodiment is included in at least one embodiment of the present application, but does not mean that the specific part, structure, material, or feature is included in all embodiments.

Therefore, the phrase "in 1 embodiment" or "in one embodiment" expressed in various scenes throughout the present specification does not necessarily mean the same embodiment of the present application. Furthermore, the particular features, structures, materials, or characteristics may be combined in any suitable manner in 1 or more embodiments.

PET scanners in embodiments of the invention sometimes have different electronics configurations. Non-limiting exemplary configurations are shown in fig. 9 and 10.

In this case, the PET scanner, i.e., the entire scanner, is generally in the form of a ring.

The area is a relatively large portion of a scanner such as a quarter-segment constituted by advanced data processing, data transfer, clock control, signal processing, and the like. The scanner may also have several regions. Timing offsets/drifts based on clock distribution are sometimes area-based.

The Detector Unit (DU) is a relatively independent module, and is configured by data transfer, clock control, signal processing, and the like. One region may have 10-20 DUs. Timing offset/drift based on clock distribution is sometimes the DU body.

The circuit Board (Board) is an electronic equipment circuit Board and is composed of several signal processing circuits for channels. The DU may also have 5-20 circuit boards. Timing offsets/drifts based on power supply devices are sometimes circuit board substrates.

An ASIC is the smallest signal processing unit, typically consisting of 1 timing processing channel and several energy processing channels. The Circuit board may have 1 to 10 Application Specific Integrated Circuits (ASICs) facing a Specific Application. Timing offsets/drifts are sometimes the ASIC body.

The crystal is the smallest element in a scanner. The ASIC may also perform signal processing on the tens of crystals.

As described above, in the embodiment, the detector 300 is configured by a plurality of processing units, and the circuit (processing circuit) of the embodiment corrects the relative timing offset within the set (set) of processing units based on the data acquired via the radiation source, and then calibrates the timing offset between the sets (set) of processing units using the intrinsic radiation. Next, the circuit of the embodiment determines the total timing offset by summing the corrected relative timing offset within the group (set) of processing elements and the calibrated timing offset between the groups of processing elements, thereby calibrating the detector 300.

Here, as the "processing unit", various embodiments can be considered. As an example, as shown in fig. 9, the processing units may have a plurality of detectors of the electronic device circuit boards, respectively. That is, the circuit of the embodiment may first correct the relative timing offset within the set of detectors (set) using the radiation source, and then calibrate the timing offset between the sets of detectors using the intrinsic radiation. As another example, as shown in fig. 10, the group (set) of processing units may include a plurality of detector units in one area (for example, a relatively large portion of a scanner such as a quadrant configured by high-level data processing, data transfer, clock control, signal processing, and the like). In this case, the circuit of the embodiment may first correct the relative timing offset within one region using the radiation source, and then calibrate the timing offset between the regions using the intrinsic radiation. As another example, the processing unit of the embodiment may be an electronic device board having a plurality of ASICs. In addition, as another example, the set (set) of the processing unit of the embodiment may be one detector unit. That is, the set of processing units may be a detector unit.

Timing calibration is usually performed at different stages (stages). The number of components in each stage of the electronic device structure is expanded and reduced by approximately 1 order of magnitude. For the same statistical uncertainty, the collection time and the analysis time used to calibrate the timing offsets in each stage vary widely. For example, DU sometimes contains 500-1000 crystals, so the collection time required for calibration of DU offsets is roughly √ (500-1000) or 20-30 times shorter (ignoring calculation time) than for calibration of crystal offsets.

However, it is sometimes necessary to perform timing calibration only at a certain stage (stage). After initial timing calibration, timing drift occurs at a certain level depending on the cause of the timing drift. Therefore, maintenance timing calibration needs to be performed only at a stage where timing drift occurs, and thus is sometimes performed very quickly.

Timing walk can be corrected by including an energy-dependent condition in the offset correction. By correcting the timing walk, a better time resolution (i.e., measuring the timing difference) can be obtainedA narrower distribution). Typically, in PET, imaging is performed using only 511keV peripheral gamma rays detected through a narrow window. Linear walk correction (i.e., linear and energy-dependent walk correction) is typically sufficient where only events in a narrow energy window are used. For example, can be represented by t_offset＝t_offset(E＝511)+W₁(E-511) writing out an offset containing a linear walk correction. In the formula, W₁Is a linear walk correction factor. The walk correction can be extended to t_offset＝t_offset(E＝511)+W₁(E-511)+W₂(E-511)²+…+W_n(E-511)ⁿEtc., including non-linear conditions. In the formula, W₁To W_nIs the walk correction factor.

Here, the approach suggested in this specification uses events that span a very large amplitude energy range (in order to reduce the total collection time to a practical range) in the "intrinsic radiation" part of the calibration, thus substantially improving performance through non-linear walk correction. Since an arbitrary function used for the ambulatory correction can be expressed by taylor series expansion, this is similar to the case where f is set as a function to t_offsetThe same holds true for f (e), but the functional form of f may become nonlinear.

During an initial full timing calibration, the timing offset of each crystal and the timing walk of each crystal are calibrated. In order for each stage of the crystal to pair with more than 2 stages of the crystal within the opposing DU, the source needs to have sufficient thickness. Data using a positron source and intrinsic radiation is sometimes acquired separately or simultaneously.

In one embodiment, data is sometimes acquired separately using a positron emitting source and intrinsic radiation. The positron emitting source may be at least 1 of a Ge-68 source, a F18-FDG source, or a Na-22 source.

In particular, during data collection using a positron emitting source, a method is disclosed for centering the positron emitting source in the Field Of View (FOV) Of the scanner detector, and acquiring coincidence data using the positron emitting source. The FOV is sometimes collected using standard clinical data and the time window is counted simultaneously. The number of simultaneous counting events using a positron emitting source needs to be a number sufficient to be calibrated from the peak positions of the TOF difference histograms for the individual crystals.

During data collection using intrinsic radiation, a method is disclosed in which all radiation sources from the scanner are removed and coincidence data is acquired using intrinsic radiation. The FOV is sometimes collected using standard clinical data. In order to cope with the movement of radiation particles such as gamma particles in the entire scanner, the counting time window needs to be sufficiently large. The number of simultaneous counting events using intrinsic radiation needs to be a number sufficient to be calibrated from the peak position of the TOF difference histogram associated with each DU.

In yet another embodiment, data is sometimes acquired simultaneously using a positron emitting source and intrinsic radiation.

In particular, methods are disclosed for acquiring coincidence count data using a positron emitting source and intrinsic radiation. The FOV is sometimes collected using standard clinical data. To cope with the movement of the radiation particles over the entire scanner, the coincidence counting time window needs to be sufficiently large. The number of simultaneous counting events using a positron emitting source needs to be a number sufficient to be calibrated from the peak position of the TOF difference histogram for each crystal, and the number of simultaneous counting events using intrinsic radiation needs to be a number sufficient to be calibrated from the peak position of the TOF difference histogram for each DU.

When data using a positron emission source event and an intrinsic emission event are acquired simultaneously during data analysis, they may be separated by a TOF difference. The FOV can also be used to separate the source events from the intrinsic radiation events when the source events are focused on a narrow FOV while the intrinsic radiation events have a wider effective range.

During data analysis, the timing correction is divided into 3 different factors, namely, the non-energy-dependent relative timing shift for each crystal in opposing DU pairs, the timing drift correction coefficient for each crystal, and the non-energy-dependent timing shift between DUs. The non-energy-dependent relative timing offset for each crystal in opposing DU pairs, and the timing walk correction factor for each crystal, are calculated from the positron radiation source data, but the non-energy-dependent timing offset between DUs is calculated from the intrinsic radiation data. In the following description, the relative timing offset within opposing DU pairs and the timing offset between DUs mean the energy-independent condition.

For timing calibration within opposing DU pairs, the method is to segment coincidence data using a positron emitting source into N/2DU pairs for a PET scanner having N DUs. The relative timing offset and timing walk correction coefficients within opposing DU pairs are sometimes calibrated simultaneously for different DU pairs. In the case where the source is not perfectly centered, the annihilation position correction is applied to the TOF difference for all events. The relative timing offset of each crystal in the opposing DU pair is repeatedly calculated by: i) calculating a timing offset by finding a peak position of a timing histogram for each crystal; ii) correcting the TOF difference for the above calibrated timing offset of each crystal, then repeating steps i) and ii) until the sequence converges; iii) the final timing offset of each crystal within the DU pair is the sum of the timing offsets of each crystal calibrated by all repetitions.

After correction of the relative timing offset of each crystal in the opposing DU pair, i) plot the timing versus energy curve for each particular crystal taking into account the LOR associated with that particular crystal with any crystal on the opposite side; ii) applying a suitable fit (e.g., a linear fit or an exponential function fit) to the timing-energy curve of the crystal, thereby enabling calculation of the walk correction factor for each crystal, calculating the timing walk correction factor for each crystal.

Regarding the timing offset calibration between DUs, the timing offset correction and the timing walk correction within DU pairs are applied to the radiation coincidence counting data. For event location correction of TOF differences, it is not necessary since the TOF difference histograms of DU pairs are symmetric. However, to achieve a narrower timing histogram, event location correction can be applied to TOF differences for all events. The timing offset of each DU can be repeatedly calculated by the following process: i) calculating a timing offset by finding a peak position of a timing histogram for each DU; ii) correcting the TOF difference for the above calibrated timing offset of each DU, and then repeating steps i) and ii) until the sequence converges; iii) the last timing offset of each DU is the sum of the timing offsets of each DU calibrated by all repetitions.

In another embodiment, the timing offset for each DU is sometimes computed analytically. In particular, TOF difference histograms are calculated for each DU pair covered by the data collection FOV. The timing center for each DU pair is calculated by finding the peak position of the TOF difference histogram. The set of equations (set) is sometimes formed from the timing center for each DU pair. The variable is the timing offset of each DU. The rank (rank) of the coefficient matrix of the equation needs to be equal to the number of DUs. The timing offset for each DU can be calculated by solving the above equation. The timing offset of each DU can be calculated from the timing center of each DU pair using a neural Network (neural Network).

In yet another embodiment, the timing offset for each DU can be calculated using a neural network. In particular, the input to the neural network (for example) may be an arrangement in which each column corresponds to a timing histogram for a single DU. The output is an offset for each DU. The neural network can be trained using target offset data generated using any conventional timing offset calibration method. Training requires data from multiple systems. When a network must be trained in a state where only a small number of systems can be constructed, a large number of additional training data sets may be generated using data enhancement (described later).

In particular, data enhancement is performed by acquiring data from any existing system (e.g., 3 to 4), calibrating each system using conventional timing offset calibration, generating a corrected timing histogram for each DU pair using the calibration, generating a random timing offset for each DU to realize a plurality of systems (hundreds or thousands), and applying the random timing offset to the corrected timing histogram to construct an enhanced data set for the DU in the realization of each system. The target offsets associated with these enhancement data sets are known offsets derived from the random timing offsets generated for each DU.

Neural network designs are sometimes (to reduce the number of parameters required) convolutional neural networks. In this case, the convolutional layer is one-dimensional that functions only in a histogram from a single DU (for example, in the case where each column corresponds to a histogram from a single DU as described above, the convolutional layer is a column of the input matrix).

By the disclosed method, short time timing calibration using intrinsic radiation is enabled.

In particular, a timing offset is calculated for each processing unit. In this specification, the processing unit may be a DU or an electronic device processing unit within a DU. Data collection using intrinsic radiation is performed to remove all radiation sources from the scanner and acquire coincidence data using intrinsic radiation. The FOV is sometimes collected using standard clinical data. To cope with the movement of the radiation particles over the whole scanner, the coincidence counting time window must be sufficiently large. The number of simultaneous counting events using intrinsic radiation needs to be a number sufficient to be calibrated from the peak position of the TOF histogram for each processing unit.

In data analysis, timing offset correction and timing walk correction from initial timing calibration are applied before short-time timing calibration is performed. The data analysis order for calibrating the timing offset for each processing unit in the short-time timing calibration is the same as the data analysis order for calibrating the timing offset for each DU in the initial full timing calibration.

In another embodiment, short time timing calibrations using intrinsic radiation and positron radiation sources can be performed by the disclosed method. When calibration is required for timing offsets for processing units smaller than DU, a short time timing calibration can be calculated as in the initial timing calibration.

With respect to data collection, the number of coincidence counting events using positron emitting sources needs to be a number sufficient to be calibrated from the peak positions of the TOF difference histograms for the respective processing units, and the number of coincidence counting events using intrinsic radiation is the same as the initial timing calibration, except that it needs to be a number sufficient to be calibrated from the peak positions of the TOF histograms for the respective processing units.

Data interpretation is similar to initial timing calibration.

Timing offset correction and timing walk correction from the initial timing calibration are applied before the timing calibration is performed for a short time. The timing offset for each processing unit in the opposing DU pair is calculated in the central data using a positron emitting source, as in the initial full timing calibration. The timing offset between DUs is calibrated using intrinsic radiation as in the initial full timing calibration.

Fig. 12A and 12B show an example of TOF difference histograms of DU pairs from lutetium background radiation data when calculating timing offsets between DUs. The TOF difference is calculated as (timestamp of the first hit-timestamp of the second hit).

The timing center of the DU to TOF difference histogram can be found by gaussian fitting to a full curve, or parabolic fitting to the peak region. The timing centers of the TOF difference histograms of DU pairs can also be found using Neural Networks (NN).

As an equation for determining the timing offset of each DU, the following equation holds.

Tcenter_14-33＝Toffset₁₄–Toffset₃₃-Tdiff_distance

Tcenter_33-14＝Toffset₃₃–Toffset₁₄-Tdiff_distance

Toffset₁₄–Toffset₃₃＝(Tcenter_14-33–Tcenter_33-14)/2

Here, Tdiff_distanceThe event position correction for the TOF difference cancels out when the timing offset difference between DU14 and DU33 is calculated.

Fig. 13 shows another mode of calculation of TOF difference histograms of DU pairs from lutetium background radiation data when calculating timing offsets between DUs. The TOF difference is calculated as (timestamp of DU 14-timestamp of DU 33). The timing center of the DU to TOF difference histogram can be found by gaussian fitting to a full curve, or parabolic fitting to the peak region. The timing centers of the DU to TOF difference histograms can also be found using Neural Networks (NN).

As an equation of the timing offset for each DU, the following equation holds.

Tcenter_left＝Toffset₁₄–Toffset₃₃–Tdiff_distance

Tcenter_right＝Toffset₁₄–Toffset₃₃+Tdiff_distance

Toffset₁₄–Toffset₃₃＝(Tcenter_right+Tcenter_left)/2

Fig. 14 shows an example of a TOF difference histogram of DU pairs from lutetium background radiation data and Ge source data taken simultaneously. The TOF difference is calculated as (timestamp of DU 14-timestamp of DU 33). The positron source data and lutetium intrinsic radiation data are sometimes separated by a difference in time of flight difference (TOF).

The timing offset between the DU pairs is greatly reduced after a short time timing calibration. For example, fig. 15A shows the timing offset between DU pairs before short time timing calibration, and fig. 15B shows the timing offset between DU pairs after short time timing calibration. According to one embodiment disclosed herein, an accurate, simple, and rapid method for timing calibration of TOF PET scanners is provided.

To effectively reduce the statistical noise of reconstructed images for TOF PET scanners and improve the quality of the images, the various embodiments discussed in this specification provide good temporal resolution and are used in order to maintain accurate timing corrections between daily clinical uses, enabling images with a reduced number of artifacts for TOF PET scanners.

According to one embodiment, the timing offset calibration is performed by aggregating and pairing all processing units calibrated by the simultaneous events.

According to another embodiment, the timing offset calibration is performed by: until a sufficient number of simultaneous gamma photons between groups brings about a sufficient timing offset calibration for all of the crystals, the groups (groups) of overlapping crystals caused by simultaneous gamma photons are grouped and performed in pairs.

According to one embodiment, during an initial full timing calibration, (1) timing offsets and timing walk within the DU pairs or DUs are calibrated by deploying a limited range positron emitting source in the scanner FOV, and (2) after correction of the timing offsets and timing walk within the DU pairs or DUs, the timing offsets between the DU pairs or DUs are calibrated using intrinsic radiation (e.g., lutetium).

According to one embodiment, in step (1), the limited-range radioactive source preferably has a thickness sufficient for the crystal to pair with multiple crystals within other DUs.

According to another embodiment, during each day of clinical use, (1) timing offset corrections from initial timing calibrations and timing walk corrections are applied prior to timing calibrations performed during each day of clinical use; (2) timing offsets for each processing unit are calculated using intrinsic radiation (e.g., lutetium) with a limited range positron emission source within the scanner FOV; and/or (3) the timing offset of each processing element is calibrated during periods when the scanner is not in use, using intrinsic radiation (e.g., lutetium).

In 2 different implementations, (1) data using a limited range positron emission source and intrinsic radiation (e.g., lutetium) are acquired separately; (2) data is acquired simultaneously using a limited range positron emission source and intrinsic radiation (e.g., lutetium).

Advantageously, if at least the embodiments disclosed herein are used, (1) there is no need to move the source during initial full-timing calibration, or to use a large limited-range source; (2) the ability to perform short time timing calibrations during daily clinical use without the use of an external radiation source; (3) the simplified method does not require position dependent timing correction; (4) the calibration is performed relatively fast due to the simultaneous processing and the simple method.

The methods and systems described herein can be implemented in some technologies, but can be generally adapted to processing circuitry for performing the calibration described herein. In one embodiment, the processing circuitry is implemented alone or in combination with an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), a general purpose Array of Logic (GAL), a Programmable Logic Array (PAL), a Programmable circuit of limited to 1 degree of Logic gates (e.g., using fuses), or a combination of reprogrammable Logic gates.

Further, the processing circuitry comprises a computer processor with embedded and/or external non-volatile computer readable memory (e.g., RAM, SRAM, FRAM, PROM, EPROM and/or EEPROM) storing a computer program (binary executable commands and/or interpreted computer commands) for controlling the computer processor to perform the processes described in this specification. Computer processor circuits sometimes support single-threaded or multi-threaded, respectively, and mount a single processor or multiple processors, each having a single core or multiple cores.

In one embodiment using a neural network, the processing circuitry used to train the artificial neural network need not be the same as the processing circuitry used to install the trained artificial neural network that performs the calibration described herein. For example, the processing circuitry and memory are used to fabricate a trained artificial neural network (e.g., defined by interconnections and weights), and the FPGA may also be used to install the trained artificial neural network. In addition, in order to improve training and use of the trained artificial neural network, a serial mounting method or a parallel mounting method (for example, by installing the neural network trained in a parallel processor architecture such as a graphics processor architecture) may be used in order to improve performance.

According to at least one embodiment described above, calibration with good accuracy can be performed.

Several embodiments have been described, but these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in other various manners, and various omissions, substitutions, changes, and combinations of the embodiments can be made without departing from the scope of the invention. These embodiments and modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalent scope thereof.

The above embodiments disclose the following remarks as one aspect and optional features of the invention.

(attached note 1)

In one aspect of the present invention, a PET apparatus is provided with a detector and a circuit. The detector detects a pair of coincident events generated by annihilation of positrons emitted from a radiation source disposed in a field of view (FOV) of the detector, and is configured by a plurality of processing units. A circuit corrects relative timing offsets within the set (set) of processing units based on data acquired via the radiation source, corrects timing offsets between the set (set) of processing units using intrinsic radiation, and determines a total timing offset based on the corrected relative timing offsets within the set (set) of processing units and the corrected timing offsets between the set (set) of processing units, thereby performing calibration of the detector.

(attached note 2)

The circuit may determine a total timing offset as a sum of the relative timing offset and the calibrated timing offset, and thereby perform calibration of the detector.

(attached note 3)

The circuit may further correct the timing movement within the set (set) of processing units based on data acquired via the radiation source.

(attached note 4)

The timed walk may also be a non-linear timed walk.

(attached note 5)

Alternatively, the intrinsic radiation may be radiation generated by decay of a radioactive material that is part of a scintillator array of the PET apparatus.

(attached note 6)

Alternatively, the decay process of the intrinsic radiation may comprise at least 2 substantially simultaneous emissions resulting from the coincidence counting phenomenon as a result of the decay process.

(attached note 7)

Alternatively, the intrinsic radiation is radiation generated by decay of a radioactive material that is part of a scintillator array of the PET apparatus,

the decay process of the intrinsic radiation contains the radiation, sometimes compton scattered, within the detector that causes the counting phenomenon due to the radiation.

(attached note 8)

The intrinsic radiation may be radiation emitted from at least one of the scintillator, an adhesive holding the reflector in place, the reflector itself, or the detector housing.

(attached note 9)

Alternatively, the intrinsic radiation may be Lu-176 or Co-60.

(attached note 10)

The radiation source may be a finite range annihilation radiation source in the field of view of the detector, and the finite range annihilation radiation source may include a finite range radiation source in a range in which each crystal in the scanner and a plurality of crystals in at least one other set (set) of processing units form a pair.

(attached note 11)

The radiation source may be an annihilation radiation source of limited range in the field of view of the detector,

the narrowest cross-sectional extent of the limited range annihilation radiation source is less than 10 mm.

(attached note 12)

The limited-range annihilation radiation source may be a radiation source.

(attached note 13)

It is also possible that the limited-range annihilation radiation source is a positron-emitting source.

(attached note 14)

Alternatively, the circuitry uses a neural network to calculate the timing offset for each processing unit.

(attached note 15)

Alternatively, the limited range annihilation radiation source may be at least 1 of a Ge-68 source, a F18-FDG source, or a Na-22 source.

(subsidiary 16)

It is also possible that the set of processing elements is a pair of processing elements.

(attached note 17)

The set of processing units may be at least one of a pair of simultaneous processing units and a set of processing units, which are paired according to a coincidence count via the radiation source.

(attached note 18)

The circuit may acquire a part of data from each of the intrinsic radiation and the radiation source.

(attached note 19)

The circuit may acquire a part of data from the intrinsic radiation and the radiation source at the same time.

(attached note 20)

The processing unit may be a detector unit having a plurality of electronic device circuit boards, respectively.

(attached note 21)

The set of processing elements may be a plurality of detector elements comprised by one area.

(attached note 22)

It is also possible that the set of processing units (set) is one detector unit.

(attached note 23)

The processing unit may be an electronic device circuit board having a plurality of ASICs.

(attached note 24)

In a calibration method provided in one aspect Of the present invention, a timing calibration method is performed in Time Of Flight (TOF) Positron Emission Tomography (PET), wherein a radiation source is arranged in the FOV Of a PET scanner, a relative timing offset in a set (set) Of processing units is acquired, the relative timing offset in the set (set) Of processing units is corrected, a timing offset between the sets (set) Of processing units is calibrated using intrinsic radiation, and calibration Of a detector is performed based on the corrected relative timing offset in the set (set) Of processing units and the calibrated timing offset between the sets (set) Of processing units.

Claims (15)

1. A Positron Emission Tomography (PET) apparatus includes:

a detector configured by a plurality of processing units to detect a pair of coincident events generated by annihilation of positrons emitted from a radiation source disposed in a field of view (FOV) of the detector; and

a circuit for correcting relative timing offsets within the group of the processing units based on data acquired via the radiation source, correcting timing offsets between the groups of the processing units using intrinsic radiation, and determining a total timing offset based on the corrected relative timing offsets within the group of the processing units and the corrected timing offsets between the groups of the processing units, thereby performing calibration of the detector.

2. The PET device of claim 1,

the circuit performs calibration of the detector by determining a total timing offset as a sum of the relative timing offset and the calibrated timing offset.

3. The PET device of claim 1,

the circuit further corrects the timing walk within the group of processing units based on data acquired via the radiation source.

4. The PET device of claim 1,

the intrinsic radiation is radiation generated by decay of radioactive material that is part of the scintillator array of the PET apparatus.

5. The PET device of claim 1,

the source is a limited range annihilation source within the field of view of the detector,

the limited-range annihilation radiation source includes a limited-range radiation source having a range in which each crystal in the PET device is paired with a plurality of crystals in at least one other group of the processing units.

6. The PET device of claim 1,

the source is a limited range annihilation source within the field of view of the detector,

the narrowest cross-sectional extent of the limited range annihilation radiation source is less than 10 mm.

7. The PET device of claim 1,

the circuitry uses a neural network to calculate the timing offset for each processing unit.

8. The PET device of claim 1,

the set of processing units is at least one of a pair of coincident processing units, and a set of processing units that are paired according to a coincidence count via the radiation source.

9. The PET device of claim 1,

the circuit acquires a part of data from the intrinsic radiation and the radiation source, respectively.

10. The PET device of claim 1,

the circuit simultaneously acquires a portion of data from the intrinsic radiation and the radiation source.

11. The PET device according to any one of claims 1 to 10,

the processing units are detector units each having a plurality of electronic device circuit boards.

12. The PET device according to any one of claims 1 to 10,

the set of processing units is a plurality of detector units comprised by an area.

13. The PET device according to any one of claims 1 to 10,

the group of processing units is one detector unit.

14. The PET device according to any one of claims 1 to 10,

the processing unit is an electronic device circuit board having a plurality of ASICs, respectively.

15. A calibration method for performing timing calibration in Time Of Flight (TOF) Positron Emission Tomography (PET),

a radiation source is arranged in a field of view (FOV) of a detector of a PET scanner to obtain a relative timing offset within a group of processing units, the relative timing offset within the group of processing units is corrected, a timing offset between the groups of processing units is calibrated by using intrinsic radiation, and the detector is calibrated based on the corrected relative timing offset within the group of processing units and the calibrated timing offset between the groups of processing units.

Images