(19)
(11)EP 3 073 292 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
29.04.2020 Bulletin 2020/18

(21)Application number: 13897953.9

(22)Date of filing:  25.12.2013
(51)International Patent Classification (IPC): 
G01T 7/00(2006.01)
G01T 1/36(2006.01)
G01T 1/20(2006.01)
G01T 1/29(2006.01)
(86)International application number:
PCT/CN2013/090389
(87)International publication number:
WO 2015/074312 (28.05.2015 Gazette  2015/21)

(54)

METHOD AND SYSTEM FOR RECOVERING SCINTILLATION PULSE INFORMATION

VERFAHREN UND SYSTEM ZUR WIEDERHERSTELLUNG VON SZINTILLATIONIMPULSINFORMATIONEN

PROCÉDÉ ET SYSTÈME PERMETTANT DE RECUEILLIR DES INFORMATIONS D'IMPULSION DE SCINTILLATION


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 19.11.2013 CN 201310585902

(43)Date of publication of application:
28.09.2016 Bulletin 2016/39

(73)Proprietor: Raycan Technology Co., Ltd. (Su Zhou)
Suzhou, Jiangsu 215163 (CN)

(72)Inventors:
  • XIE, Qingguo
    Suzhou, Jiangsu 215163 (CN)
  • DENG, Zhenzhou
    Suzhou, Jiangsu 215163 (CN)

(74)Representative: Kolster Oy Ab 
(Salmisaarenaukio 1) P.O. Box 204
00181 Helsinki
00181 Helsinki (FI)


(56)References cited: : 
CN-A- 101 978 289
CN-A- 102 262 238
CN-A- 102 073 059
  
  • DENG ZHENZHOU ET AL: "Empirical Bayesian energy estimation for multi-voltage threshold digitizer in PET", 2013 IEEE NUCLEAR SCIENCE SYMPOSIUM AND MEDICAL IMAGING CONFERENCE (2013 NSS/MIC), IEEE, 27 October 2013 (2013-10-27), pages 1-5, XP032601158, DOI: 10.1109/NSSMIC.2013.6829196 [retrieved on 2014-06-10]
  • ZHENZHOU DENG ET AL: "Scintillation event energy measurement via a pulse model based iterative deconvolution method", PHYSICS IN MEDICINE AND BIOLOGY, INSTITUTE OF PHYSICS PUBLISHING, BRISTOL GB, vol. 58, no. 21, 21 October 2013 (2013-10-21), pages 7815-7827, XP020252324, ISSN: 0031-9155, DOI: 10.1088/0031-9155/58/21/7815 [retrieved on 2013-10-21]
  • DENG, ZHENZHOU ET AL.: 'Empirical Bayesian Energy Estimation for Multi-Voltage Threshold Digitizer in PET' IEEE 02 November 2013, page 1, XP032601158
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

FIELD



[0001] The present disclosure relates to the field of digital signal processing, photoelectric signal processing and radiation detection, and in particular to a method and a system for recovering scintillation pulse information.

BACKGROUND



[0002] In the nuclear analysis field including positron lifetime spectrometers and positron angle-momentum association analyzers, in the nuclear detection field including double-compliance energetic particle discriminators and in the medical imaging field including the Positron Emission Tomography (abbreviated as PET hereinafter), a main working principle of detecting energetic particles is to convert a high-energy ray into an electric signal and then obtain various information of a particle event with a method of the fast electronics. In some applications of meters involving coincidence logical operations and time stamps, shaping a scintillation pulse, which slows down the process, may affect a time performance of the scintillation pulse and may increase a probability of stacking event pulses. In this case, it is more desirable to directly digitize the scintillation pulse.

[0003] In a complete digitization of a scintillation pulse, an Analog to Digital Converter (abbreviated as ADC hereinafter) with a high bandwidth and a high sampling rate is adopted to directly sample and quantize the scintillation pulse. Such digitization method cannot meet an actual requirement of multi-channel systems due to a high cost. And the above digitization solution may be simplified in two manners to reduce the cost. In one manner, the sampling rate is reduced. In this case, due to a low sampling frequency, a high-frequency component of a non-shaped scintillation pulse signal not only cannot be sampled, but also affects accuracy of a component around a Nyquist frequency due to aliasing in the frequency domain. In order to alleviate the above defect, a shaping process is added between a process of outputting the scintillation pulse and a process of digitizing the scintillation pulse. In the shaping process, a component of the scintillation pulse having a frequency higher than the Nyquist frequency is attenuated and a low-frequency component of the scintillation pulse is reserved, to reduce the aliasing in frequencies due to the digitizing of the scintillation pulse. However, it is impossible to sample a component having a frequency higher than the Nyquist frequency due to a limitation by the Nyquist frequency itself.

[0004] Besides the manner in which the ADC is simplified in the time axis, in another manner, the ADC is simplified in the voltage axis. An effective bandwidth of the ADC simplified in the voltage axis may be improved by using an open-loop designation, which is favorable to a high-speed signal processing. For example, a digitization method involving several comparators/ADC units and an ADC with interleaving open-circuit time all perform in this manner. Such ADC is characterized by a high sampling rate and a high bandwidth but a limited quantization accuracy. Since the scintillation pulse has distinctive priori knowledge, with which the accuracy of recovering a digitalized signal may be improved and a quantization level may be set optimally.

[0005] Therefore, for the above sparse quantization level ADC, it is desired to provide a new method and a new system for recovering scintillation pulse information to address issues existing in the conventional technology. An example of such technology is given in the document: Deng Zhenzhou et al.: "Empirical Bayesian energy estimation for multi-voltage threshold digitizer in PET", 2013 IEEE NUCLEAR SCIENCE SYMPOSIUM AND MEDICAL IMAGING CONFERENCE, 27 October 2013, pages 1-5.

SUMMARY



[0006] In view of the above, an object of the present disclosure is to provide a method and a system for recovering scintillation pulse information, to recover a digitized signal more accurately and to acquire more event information by optimizing a setting of quantization levels.

[0007] In order to achieve the above object, the following technical solutions are provided according to the present disclosure.

[0008] A method for recovering scintillation pulse information includes:

step S 1: acquiring a scintillation pulse database of non-stacked compliance single events in a low count, and building a scintillation pulse noise model for the scintillation pulse database of the single events, where an average pulse is calculated for the scintillation pulse database of the non-stacked compliance single events, and scintillation pulse shape information is given by the average pulse, wherein the step for calculating the average pulse comprises:

step 1.1: reducing a radiation dose of a radiation source to obtain a weak source, and reducing the number of high-energy photons captured by a detector by the weak source or by adjusting a solid angle of the weak source with respect to the detector, wherein an event received by the detector is a Poisson stream having an average count rate expressed as

where mi and qi denotes respectively a dose of the weak source and a solid angle of the weak source with respect to the detector, i denotes an index of the weak source, and n denotes the number of weak sources;

step 1.2: aligning the pulses with a constant coefficient identification method or a leading edge discrimination method; and

step 1.3: averaging the aligned pulses;

step S2: calculating a posterior probability logarithm value for a given energy value based on the scintillation pulse noise model, including:

step 2.1: loading a scintillation pulse segment S0, and calculating a likelihood function of the given energy value based on the pulse noise model, where the scintillation pulse segment starts at a time origin t0 at which the scintillation pulse segment passes upwards a threshold Vt and the scintillation pulse segment ends at a time point t0 + Δt, with Δt being greater than two times of a time constant of a falling edge of a scintillation crystal; and

step 2.2: calculating logarithms for all of the time points and adding up the logarithms, to obtain a value of a function having same monotonicity as a posterior probability function; and

step S3: performing step S2 repeatedly to obtain an energy value meeting a maximal posterior probability condition, including:

step 3.1: calculating posterior probabilities for different tentative energy values by repeatedly performing step S2 as a module;

step 3.2: searching linearly for the energy value meeting the maximal posterior probability condition; and

step 3.3: after the energy value is obtained, correcting the time origin for step 2.1 based on the energy value and repeating step 3.1 and step 3.2.



[0009] Preferably, in the above method for recovering scintillation pulse information, the scintillation pulse database in step S1 may include 2000 or more samples.

[0010] Preferably, in the above method for recovering scintillation pulse information, the scintillation pulse noise model in step S1 may include characteristics of a scintillation pulse shape, a scintillation pulse height and a scintillation pulse noise level.

[0011] Preferably, in the above method for recovering scintillation pulse information, in step S1, after the pulses are aligned, energy values of scintillation pulses corresponding to a same energy value at a same time point may be distributed dispersedly, the distribution may be defined as a noise model of the pulses, and a parameter of the noise model may be used to calculate a posterior probability value for a given energy value.

[0012] Preferably, in the above method for recovering scintillation pulse information, step S1 may further include: setting a couple of compliance pulses selected within an energy range as a couple of single events, and pre-storing and analyzing off-line such digitized electric pulses, where an average pulse signal is obtained by aligning the couple of pulses, and energy spectrums of the single events are obtained by adding up the digitized electric pulses by taking the average pulse as a system response.

[0013] A system for recovering scintillation pulse information includes:

a noise model module configured to train pre-acquired data having a low count rate to acquire a noise model of a scintillation detector system, where the noise model represents an average pulse, a noise variance and a noise distribution skewness , and the noise model is provided to the posterior probability module;

a posterior probability module configured to calculate a posterior probability to obtain a posterior probability logarithm value for a given energy value; and

an energy value search module configured to search for an energy value meeting a maximal posterior probability condition, wherein the energy value search module comprises:

a tentative energy calculation module configured to calculate a posterior probability logarithm value for a tentative energy value by calling the posterior probability module;

a linear search module configured to calculate an energy value meeting a maximal posterior probability condition with a linear search algorithm; and

a time origin correction module configured to correct a time origin for a sample and repeatedly call the tentative energy calculation module and the linear search module.



[0014] Preferably, in the above system for recovering scintillation pulse information, the noise model module may include:

an average pulse module configured to calculate an average pulse of single events;

a pulse noise module configured to calculate a noise variance of pulses of the single events; and

a probability density distribution module configured to estimate a probability density distribution function of noise.



[0015] Preferably, in the above system for recovering scintillation pulse information, the posterior probability module may include:

a likelihood function distribution module configured to calculate a likelihood function value of each point based on the given energy value and the pulse noise model input by the noise model module; and

a time point product module configured to calculate a posterior probability logarithm value.



[0016] It can be seen from the above technical solutions that, with the above method and system for recovering scintillation pulse information according to the embodiments of the present disclosure, energy information of the respective single events can be calculated effectively, a count rate of the system in an energy window is increased, and an energy resolution of the system is improved. Therefore, the method and system are especially suitable for processing a scintillation event in an off-line environment.

[0017] It can be seen from the above technical solutions that, in the embodiments of the present disclosure, a pulse height or energy information of the scintillation pulse are estimated accurately by calculating parameters of a scintillation event corresponding to a maximal posterior estimation of a given system.

[0018] Compared with the conventional technologies, the present disclosure has the following advantages:
  1. (1) a good energy resolution;
  2. (2) a high count rate for a given energy window; and
  3. (3) an ability to avoid quantization error caused by the small number of quantization levels.

BRIEF DESCRIPTION OF THE DRAWINGS



[0019] The drawings to be used in the description of the embodiments or the conventional technology will be described briefly in the following, so that the technical solutions according to the embodiments of the present disclosure or according to the conventional technology will become clearer.

Figure 1 is a structural diagram of a system for recovering scintillation pulse information according to the present disclosure;

Figure 2 is a flowchart of a method for recovering scintillation pulse information according to the present disclosure;

Figure 3 is a schematic diagram of a typical scintillation pulse;

Figure 4 is a schematic diagram of a set of scintillation pulses having same energy;

Figure 5 shows digitization results of the scintillation pulse shown in Figure 3 under an ADC with four thresholds;

Figure 6 is a schematic diagram of an energy spectrum in a case of a poor energy resolution, in which energy is calculated with digital gated integration;

Figure 7 is a schematic diagram of an energy spectrum in a case of a poor energy resolution, in which energy is calculated with a method according to the present disclosure;

Figure 8 is a schematic diagram of an energy spectrum in a case of a poor energy resolution, in which energy is calculated with a least-square fitting algorithm;

Figure 9 is a schematic diagram of an energy spectrum in a case of a good energy resolution, in which energy is calculated with digital gated integration;

Figure 10 is a schematic diagram of an energy spectrum in a case of a good energy resolution, in which energy is calculated with a method according to the present disclosure;

Figure 11 is a schematic diagram of an energy spectrum in a case of a good energy resolution, in which energy is calculated with a least-square fitting algorithm;

Figure 12 is a scatter diagram of energy values obtained by a least-square fitting algorithm vs. energy values obtained by digital gated integration, in which the energy values are as shown in Figure 4;

Figure 13 is a scatter diagram of energy values obtained by a method according to the present disclosure vs. energy values obtained by digital gated integration, in which the energy values are as shown in Figure 4;

Figure 14 is a scatter diagram of energy values obtained by a least-square fitting algorithm vs. energy values obtained by digital gated integration, in which the energy values are as shown in Figure 7;

Figure 15 is a scatter diagram of energy values obtained by a method according to the present disclosure vs. energy values obtained by digital gated integration, in which the energy values are as shown in Figure 7;

Figure 16 is a diagram showing a process for searching for an optimal one-dimension solution for an energy value;

Figure 17 is a schematic diagram of a typical system according to the present disclosure; and

Figure 18 is a schematic diagram of another typical system according to the present disclosure.


DETAILED DESCRIPTION OF EMBODIMENTS



[0020] The present disclosure provides a method and a system for recovering scintillation pulse information. With the method and the system, pulse energy information sampled by a sparse quantization level ADC can be recovered, a height estimation of scintillation pulse can be estimated more accurately and a count rate of the system in an energy window can be increased.

[0021] The method for recovering scintillation pulse information according to the present disclosure includes the following step 1 to step 3.

[0022] In step 1, a pulse database of non-stacked compliance single events in a low count is acquired, and then a noise model is built for the pulse database of the single events. It is preferable that the pulse database includes more than 2000 samples. The noise model is more accurate with the increasing number of pulses.

[0023] In step 2, a posterior probability logarithm value for a given energy value is calculated.

[0024] In step 3, an energy value meeting a maximal posterior probability condition is searched for linearly.

[0025] The system for recovering scintillation pulse information according to the present disclosure includes a noise model module, a posterior probability module and an energy value search module. The noise model module is configured to obtain a noise model from a single event pulse set, and the other modules are configured to process scintillation pulse signals obtained by a sparse quantization level ADC. The posterior probability module is configured to calculate a posterior probability of an inputted energy value. And the energy value search module is configured to search for an energy value meeting a maximal posterior probability condition.

[0026] The technical solution according to the embodiments of the present disclosure will be described clearly and completely as follows in conjunction with the appended drawings.

[0027] In the method, the energy value meeting the maximal posterior probability condition is obtained based on the Bayes principle.

[0028] In the method for recovering scintillation pulse information according to the present disclosure, a noise model is firstly obtained based on a pre-acquired database. The noise model is used as inherent prior knowledge of an electric signal to calculate a posterior probability of a sparsely quantized scintillation pulse. And an energy value meeting a maximal posterior probability condition is searched for linearly by repeatedly calling the posterior probability module.

[0029] As shown in Figure 2, the method for recovering scintillation pulse information according to the present disclosure includes the following steps S1 to S3.

[0030] In step S1, a scintillation pulse database of non-stacked compliance single events in a low count is acquired. A scintillation pulse noise model is built for the scintillation pulse database of the single events. An average pulse is calculated for the scintillation pulse database of the non-stacked compliance single events. Scintillation pulse shape information is given based on the average pulse. It is required that the pulse database includes more than 2000 samples. And a statistic noise is smaller with the increasing number of pulses.

[0031] The scintillation pulse noise model in step S1 includes characteristics of a scintillation pulse shape, a scintillation pulse height and a scintillation pulse noise level. These characteristics are parameters for calculating a posterior probability of energy of a couple of samples at a single time point, and a posterior probability corresponding to the energy cannot be calculated without values of the above characteristics. The parameters depend on the system. A noise model can be built by those skilled in the art with different methods. Generally, the noise model may be obtained by a simulation or experimental measurement. In a case of the simulation, a distribution of voltage values at different time points of each pulse is given based on an assumption, and the distribution is the noise model. In a case of the experimental measurement, sampled scintillation pulses are aligned, thereby obtaining a distribution of voltage values at each time point (a scintillation pulse noise model).

[0032] The average pulse is calculated in step S1 with the following steps S1.1 to S1.3.

[0033] In step 1.1, a radiation dose of a radiation source is reduced to obtain a weak source. The number of high-energy photons captured by a detector is reduced by the weak source or by adjusting a solid angle of the weak source with respect to the detector. A probability of stacking is small under an extreme low count rate. An event received by a detector is a Poisson stream having an average count rate expressed as

where mi denotes a dose of the weak source and qi denotes the solid angle of the weak source to the detector, i denotes an index of the weak source, and n denotes the number of weak sources. It may be configured that n = 1 and m is small enough in acquiring the pulse database.

[0034] In step 1.2, the pulses are aligned with a constant coefficient identification method or a leading edge discrimination method.

[0035] In step 1.3, the aligned pulses are averaged.

[0036] After the pulses are aligned in step S1, energy values of scintillation pulses corresponding to a same energy value at a same time point are distributed dispersedly. The distribution is defined as a noise model of the pulses. A parameter of the noise model is used to calculate a posterior probability value for a given energy value.

[0037] Reference is made to Figure 4, which is a schematic diagram of a set of scintillation pulses having same energy. Step S1 further includes: setting a couple of compliance pulses selected in an energy range as a couple of single events, pre-storing and then analyzing off-line such digitized electric pulses. The couple of pulses are aligned to obtain an average pulse signal and a pulse residual sequence. And an energy spectrum of single event data is obtained by adding up the digital pulses by taking the average pulse as a system response.

[0038] Unless specified otherwise, the scintillation pulses according to the present disclosure each have a positive value. The energy spectrum of single event data may be obtained by adding the digital pulses up. If an actually-acquired pulse is a negative pulse, a phase inversion has been performed by default. Therefore, a rising edge of an electric pulse is its leading edge and a falling edge of the electric pulse is its tail portion. All intermediate information obtained in processing an original pulse database may be referred to as a derivative pulse database. And the derivative pulse database may be obtained by performing operations, such as filtering, interpolation, fitting, extrapolation and combination, on the original pulse database.

[0039] In step S2, a posterior probability logarithm value for a given energy value is calculated based on the scintillation pulse noise model.

[0040] An energy value meeting the maximal posterior probability condition reflects a basic principle in statistics, which is that a probability that a current observation occurs at an energy value is greater than a probability that the observation occurs at other energy values. The probability is calculated by multiplying a prior probability of an energy value by a likelihood probability that an observation occurs at the energy value.

[0041] Step S2 includes the following step 2.1 and step 2.2.

[0042] In step 2.1, a scintillation pulse segment S0 is loaded, and a likelihood function of the given energy value is calculated based on the noise model of pulse obtained above. The scintillation pulse segment starts at a time point t0 at which the scintillation pulse segment passes upwards a threshold Vt, and the scintillation pulse segment ends at a time point t0 + Δt, with Δt being greater than two times of a time constant of a falling edge of a scintillation crystal.

[0043] In step 2.2, logarithms for all of the time points are calculated and added up, to obtain a value of a function having same monotonicity as a posterior probability function.

[0044] The likelihood function does not need to be directly calculated in an actual calculation, and is obtained by calculating a logarithm value of the likelihood function. The logarithmic likelihood function is equivalent to a sum of logarithm functions of the pulse at individual time points. Since the likelihood function may have other forms, the form of the likelihood function is not fixed.

[0045] The time point means an arrival time of the pulse. And the arrival time is considered as a zero point for a time point of the pulse. Since a logarithm of a function has the same monotonicity as the function, the logarithm of the function achieves an extreme value as the function achieves an extreme value. Therefore, maximizing a certain function is equivalent to maximizing a logarithm of the function. Therefore, as performed in step S2, maximizing a logarithm of a posterior probability is equivalent to maximizing the posterior probability.

[0046] In step S3, an energy value meeting a maximal posterior probability condition is calculated by repeatedly performing step S2.

[0047] In step S3, an energy correction is performed on a time origin of the pulse after each energy calculation. The time points are corrected not only for making the energy value more accurate but also for obtaining more accurate time information. In a PET or a nuclear measurement apparatus, a single event pulse is completely represented with time information and energy information. The shape information is given by the average pulse, and is unnecessary for most of nuclear measurement apparatuses.

[0048] Step S3 includes the following step 3.1 to step 3.3.

[0049] In step 3.1, step S2 is performed repeatedly as a module, to calculate posterior probabilities corresponding to different tentative energy values.

[0050] In step 3.2, the energy value meeting the maximal posterior probability condition is searched for linearly. Existing linear-search algorithms include a golden section method, a bi-search method, a Fibonacci series method and the like. As shown in Figure 16, Figure 16 is a diagram showing a process for searching for an optimal a one-dimension solution for an energy value.

[0051] A maximized posterior probability is obtained by multiplying a maximized likelihood function by an energy probability distribution function. And a posterior probability is obtained by multiplying a likelihood function by an energy probability distribution function.

[0052] In step 3.3, after the energy value is obtained, the time origin is corrected in step 2.1 based on the energy value, and then step 3.1 and step 3.2 are repeated.

[0053] The energy value is calculated based on digitized samples, that is, the measured energy value depends on time. The time is not accurate without being corrected based on the energy value, and a "time walk" phenomenon may occur. Therefore, inaccurate time is corrected after the energy value is obtained. The energy value is calculated again after the time is corrected, and the time is corrected again after the energy value is obtained. An iteration process of time-energy-time-energy is performed to improve the accuracy of the time and the energy measurement. Therefore, the iteration related to time is introduced to improve the accuracy of the measurement.

[0054] Reference is made to Figure 6 to Figure 11, which are schematic diagrams of energy spectra obtained by adopting different energy calculation methods in a case of a poor energy resolution or in a case of a good energy resolution. The adopted energy calculation methods include a digital gated integration, a method according to the present disclosure and a least-square fitting algorithm.

[0055] Figure 6 is a schematic diagram of an energy spectrum in a case of a poor energy resolution, in which energy is calculated with the digital gated integration. The energy resolution is calculated by dividing peak position by a peak width (a full width at half maximum). In the gated integration, a gated signal is obtained by a threshold inversion of a rising edge. A pulse is integrated in such gated range, and the obtained integration value is normalized to represent an energy value. The normalization is performed by normalizing a full energy peak with respect to 511keV (kilo-electron-volt).

[0056] Figure 7 is a schematic diagram of an energy spectrum in a case of a poor energy resolution, in which energy is calculated with the method according to the present disclosure. And a used method for calculating the energy resolution and a used method for normalization are the same as those shown in Figure 6.

[0057] Figure 8 is a schematic diagram of an energy spectrum in a case of a poor energy resolution, in which energy is calculated with the least-square fitting algorithm. And a used method for calculating the energy resolution and a used method for normalization are the same as those shown in Figure 6.

[0058] Figure 9 is a schematic diagram of an energy spectrum in a case of a good energy resolution, in which energy is calculated with the digital gated integration. And a used method for calculating the energy resolution and a used method for normalization are the same as those shown in Figure 6.

[0059] Figure 10 is a schematic diagram of an energy spectrum in a case of a good energy resolution, in which energy is calculated with the method according to the present disclosure. And a used method for calculating the energy resolution and a used method for normalization are the same as those shown in Figure 6.

[0060] Figure 11 is a schematic diagram of an energy spectrum in a case of a good energy resolution, in which energy is calculated with the least-square fitting algorithm. And a used method for calculating the energy resolution and a used method for normalization are the same as those shown in Figure 6.

[0061] Reference is made to Figure 12 to Figure 15, which are scatter diagrams of energy values obtained by the least-square fitting algorithm vs. energy values obtained by the digital gated integration and scatter diagrams of energy values obtained by the method according to the present disclosure vs. energy values obtained by the digital gated integration.

[0062] Figure 12 is a scatter diagram of energy values obtained by the least-square fitting algorithm vs. energy values obtained by the digital gated integration, and the energy values are as shown in Figure 4. A horizontal coordinate in the scatter diagram represents an energy value obtained by the gated integration, and a vertical coordinate in the scatter diagram represents an energy value obtained by the pulse fitting. Taking the gated integration as a golden standard, an error level of an energy calculation method with respect to the gated integration may be used to evaluate the accuracy of the energy calculation method. Performances of the respective energy calculation methods may be observed by representing energy accuracy as a Root Mean Square Error (RMSE).

[0063] Figure 13 is a scatter diagram of energy values obtained by the method according to the present disclosure vs. the energy values obtained by the digital gated integration, and the energy values are as shown in Figure 4. A horizontal coordinate in the scatter diagram represents an energy value obtained by the gated integration, and a vertical coordinate in the scatter diagram represents an energy value obtained by the pulse fitting.

[0064] Figure 14 is a scatter diagram of energy values obtained by the least-square fitting algorithm vs. energy values obtained by the digital gated integration, and the energy values are as shown in Figure 7. A horizontal coordinate in the scatter diagram represents an energy value obtained by the gated integration and a vertical coordinate in the scatter diagram represents an energy value obtained by the method according to the present disclosure.

[0065] Figure 15 is a scatter diagram of energy values obtained by the method according to the present disclosure vs. energy values obtained by the digital gated integration, and the energy values are as shown in Figure 7. A horizontal coordinate in the scatter diagram represents an energy value obtained by the gated integration, and a vertical coordinate in the scatter diagram represents an energy value obtained by the method according to the present disclosure.

[0066] It can be seen from energy spectrum diagrams as shown in Figure 6 to Figure 11 that, the energy resolution obtained by the method according to the present disclosure is better than that obtained by the conventional pulse fitting method, and the method according to the present disclosure facilitates rejection of a scattering event. And it can be seen from scatter diagrams as shown in Figure 12 to Figure 15 that, the method according to the present disclosure has higher accuracy and a smaller error, and is more suitable for estimating a height of a pulse and for designing some position-sensitive detector. The searching process shown in Figure 16 is a process for reducing an energy range defined in the method according to the present disclosure, which is shown in the schematic diagram as that the range is shrunk by 0.618 times each time.

[0067] With the method and system for recovering scintillation pulse information, system energy calculation accuracy is improved effectively. The method and the system are especially suitable for calculating energy by a sparse quantization level ADC digital nuclear instrument.

[0068] As shown in Figure 1, a system for recovering scintillation pulse information according to the present disclosure includes a noise model module 100, a posterior probability module 200 and an energy value search module 300. The noise model module 100 is configured to train pre-acquired data having a low count rate to obtain a noise model of a scintillation detector system. The posterior probability module 200 is configured to calculate a posterior probability to obtain a posterior probability logarithm value for a given energy value. And the energy value search module 300 is configured to search for an energy value meeting a maximal posterior probability condition.

[0069] As further shown in Figure 1, the noise model module 100 is configured to train the pre-acquired data to obtain an average pulse, a noise variance and a noise distribution skewness. And the noise model is provided to the module 200.

[0070] The noise model module 100 may include three sub-modules, i.e., an average pulse module 110, a pulse noise module 120 and a probability density distribution module 130. The average pulse module 110 is configured to calculate an average pulse of single events. The pulse noise module 120 is configured to calculate a noise variance of pulses of the single events. And the probability density distribution module 130 is configured to estimate a probability density distribution function of noise.

[0071] The posterior probability module 200 is configured to calculate a posterior probability logarithm value for a given energy value.

[0072] The posterior probability module 200 may include two sub-modules, i.e., a likelihood function distribution module 210 and a time point product module 220. The likelihood function distribution module 210 is configured to calculate a value of a likelihood function for each point based on the given energy value and the pulse noise model inputted by the module 100. And the time point product module 220 is configured to calculate a posterior probability logarithm value.

[0073] The energy value search module 300 is configured to calculate an energy value meeting a maximal posterior probability condition.

[0074] The energy value search module 300 may include three sub-modules, i.e., a tentative energy calculation module 310, a linear search module 320 and a time origin correction module 330. The tentative energy calculation module 310 is configured to calculate a posterior probability logarithm value for a tentative energy value by calling the module 200. The linear search module 320 is configured to calculate an energy value meeting the maximal posterior probability condition with a linear search algorithm. And the time origin correction module 330 is configured to correct a time origin of a sample and call the module 310 and the module 320 repeatedly.

[0075] Reference is made to Figure 3 to Figure 5. Figure 3 is a schematic diagram of a typical scintillation pulse according to the present disclosure, Figure 4 is a schematic diagram of a set of scintillation pulses having same energy, and Figure 5 shows digitization results of the scintillation pulse shown in Figure 3 under an ADC having four thresholds. In Figure 5, VI, V2, V3 and V4 represent reference voltages 1, 2, 3, 4, respectively. Reference is made to Figure 17 and Figure 18. Figure 17 is a schematic diagram of a typical system with a compliance working mode according to the present disclosure, and Figure 18 is a schematic diagram of another typical system with single channel working mode according to the present disclosure. In Figures 17 and 18, a scintillation pulse is represented as 400, a radiation source is represented as 500, a photomultiplier tube is represented as 600 and a digital oscilloscope is represented as 700. The method and the system for recovering scintillation pulse information according to the present disclosure are described in detail with several embodiments in conjunction with Figures 3 to 5 and Figures 17 and 18. In the method and the system for recovering the scintillation pulse according to the present disclosure, involved parameters and filter designing are adjusted based on a feature of the pre-acquired data, to achieve a good energy resolution performance and a short pulse duration. In the following, parameters for processing data in related application embodiments are listed.

First embodiment



[0076] In the following, parameters for processing data according to the embodiment are listed.

[0077] An actual system adopted in step S1 includes an LYSO crystal and a Hamamatsu R9800 PMT. The crystal has a size of 16.5mm×16.5mm×10.0mm. A coupling surface between the crystal and the PMT is the 100 surface, and surfaces other than the coupling surface are packaged with a Teflon adhesive tape. A data acquisition system has a sapling rate of 50GHz and a bandwidth of 16GHz. As shown in Figure 10, a radiation source includes positron annihilation gamma photons of 511kev, an average pulse rising time duration is about 2ns, and a time constant is 42.5497ns after an exponential fitting is performed on a falling edge.

[0078] Δt adopted in step 2.1 is 200ns.

Second embodiment



[0079] In the following, parameters for processing data according to the embodiment are listed.

[0080] As shown in Figure 11, an actual system adopted in step S1 includes an LYSO crystal and an FM300035 SIPM. The crystal has a size of 2.0mm×2.0mm×10.0mm. A coupling surface between the crystal and the PMT is the 100 surface, and surfaces other than the coupling surface are packaged with a Teflon adhesive tape. A data acquisition system has a sampling rate of 50GHz and a bandwidth of 16GHz. As shown in Figure 10, a radiation source includes positron annihilation gamma photons of 511kev and an average pulse rising time duration is about 5ns.

[0081] Δt adopted in step 2.1 is 300ns.

Third embodiment



[0082] In the following, parameters for processing data according to the embodiment are listed.

[0083] An actual system adopted in step S1 includes an LaBr crystal and an Hamamatsu R9800 PMT. The crystal has a size of 3.5mm×3.5mm×5.0mm. A coupling surface between the crystal and the PMT is the 100 surface, and surfaces other than the coupling surface are sealed with metal. A data acquisition system has a sampling rate of 50GHz and a bandwidth of 16GHz. As shown in Figure 10, a radiation source includes positron annihilation gamma photons of 511kev and an average pulse rising time duration is about 2ns.

[0084] Δt adopted in step 2.1 is 100ns.

[0085] The method and the system for recovering scintillation pulse information according to the present disclosure may be applied to a nuclear detection instrument, a nuclear analysis instrument and a nuclear medical instrument, in which a signal is required to be digitized directly.

[0086] The scope of the present disclosure is defined by the appended claims rather than by the foregoing description, and therefore includes all changes which come within the meaning of the claims. And the reference signs in the claims should not be construed as limiting the claim concerned.


Claims

1. A method for recovering scintillation pulse information, comprising:

step S1: acquiring a scintillation pulse database of non-stacked compliance single events in a low count, and building a scintillation pulse noise model for the scintillation pulse database of the single events, wherein an average pulse is calculated for the scintillation pulse database of the non-stacked compliance single events, and scintillation pulse shape information is given by the average pulse, wherein the step for calculating the average pulse comprises:

step 1.1: reducing a radiation dose of a radiation source to obtain a weak source, and reducing the number of high-energy photons captured by a detector by the weak source or by adjusting a solid angle of the weak source with respect to the detector, wherein an event received by the detector is a Poisson stream having an average count rate expressed as

where mi and qi denotes respectively a dose of the weak source and a solid angle of the weak source with respect to the detector, i denotes an index of the weak source, and n denotes the number of weak sources;

step 1.2: aligning the pulses with a constant coefficient identification method or a leading edge discrimination method; and

step 1.3: averaging the aligned pulses;

step S2: calculating a posterior probability logarithm value for a given energy value based on the scintillation pulse noise model, comprising:

step 2.1: loading a scintillation pulse segment S0, and calculating a likelihood function of the given energy value based on the pulse noise model, wherein the scintillation pulse segment starts at a time origin t0 at which the scintillation pulse segment passes upwards a threshold Vt, and the scintillation pulse segment ends at a time point t0 + Δt, with Δt being greater than two times of a time constant of a falling edge of a scintillation crystal; and

step 2.2: calculating logarithms for all of the time points and adding up the logarithms, to obtain a value of a function having same monotonicity as a posterior probability function; and

step S3: performing step S2 repeatedly to obtain an energy value meeting a maximal posterior probability condition, comprising:

step 3.1: calculating posterior probabilities for different tentative energy values by repeatedly performing step S2 as a module;

step 3.2: searching linearly for the energy value meeting the maximal posterior probability condition; and
characterized in that it comprises:
step 3.3: after the energy value is obtained, correcting the time origin for step 2.1 based on the energy value and repeating step 3.1 and step 3.2.


 
2. The method for recovering scintillation pulse information according to claim 1, wherein the scintillation pulse database in step S1 comprises 2000 or more samples.
 
3. The method for recovering scintillation pulse information according to claim 1, wherein the scintillation pulse noise model in step S1 comprises characteristics of a scintillation pulse shape, a scintillation pulse height and a scintillation pulse noise level.
 
4. The method for recovering scintillation pulse information according to claim 1, wherein, in step S1, after the pulses are aligned, energy values of scintillation pulses corresponding to a same energy value at a same time point are distributed dispersedly, the distribution is defined as a noise model of the pulses, and a parameter of the noise model is used to calculate a posterior probability value for a given energy value.
 
5. The method for recovering scintillation pulse information according to claim 1, wherein step S1 further comprises:

setting a couple of compliance pulses selected within an energy range as a couple of single events, and

pre-storing and then analyzing off-line such digitized electric pulses, wherein an average pulse signal is obtained by aligning the couple of pulses, and energy spectrums of the single events are obtained by adding up the digitized electric pulses by taking the average pulse as a system response.


 
6. A system for recovering scintillation pulse information, comprising:

a noise model module (100) configured to train pre-acquired data having a low count rate to acquire a noise model of a scintillation detector system, wherein the noise model represents an average pulse, a noise variance and a noise distribution skewness , and the noise model is provided to the posterior probability module;

a posterior probability module (200) configured to calculate a posterior probability to obtain a posterior probability logarithm value for a given energy value; and

an energy value search module (300) configured to search for an energy value meeting a maximal posterior probability condition, wherein the energy value search module comprises:

a tentative energy calculation module (310) configured to calculate a posterior probability logarithm value for a tentative energy value by calling the posterior probability module;

a linear search module (320) configured to calculate an energy value meeting a maximal posterior probability condition with a linear search algorithm; and characterized in that it comprises:
a time origin correction module (330) configured to correct a time origin for a sample and repeatedly call the tentative energy calculation module and the linear search module.


 
7. The system for recovering scintillation pulse information according to claim 6, wherein the noise model module comprises:

an average pulse module configured to calculate an average pulse of single events;

a pulse noise module configured to calculate a noise variance of pulses of the single events; and

a probability density distribution module configured to estimate a probability density distribution function of noise.


 
8. The system for recovering scintillation pulse information according to claim 6, wherein the posterior probability module comprises:

a likelihood function distribution module configured to calculate a likelihood function value of each point based on the given energy value and the pulse noise model input by the noise model module; and

a time point product module configured to calculate a posterior probability logarithm value.


 


Ansprüche

1. Verfahren zum Wiederherstellen von Szintillationsimpulsinformationen, das Folgendes umfasst:

Schritt S1: Erfassen einer Szintillationsimpulsdatenbank nicht gestapelter einzelner Konformitätsereignisse in einer niedrigen Zählung und Aufbauen eines Szintillationsimpulsrauschmodells für die Szintillationsimpulsdatenbank der einzelnen Ereignisse, wobei für die Szintillationsimpulsdatenbank der nicht gestapelten einzelnen Konformitätsereignisse ein mittlerer Impuls berechnet wird und Szintillationsimpulsforminformationen durch den mittleren Impuls angegeben werden, wobei der Schritt zum Berechnen des mittleren Impulses Folgendes umfasst:

Schritt 1.1: Reduzieren einer Strahlungsdosis einer Strahlungsquelle, um eine schwache Quelle zu erhalten, und Reduzieren der Anzahl von Hochenergiephotonen, die von einem Detektor aufgenommen werden, durch die schwache Quelle oder durch Anpassen eines Raumwinkels der schwachen Quelle mit Bezug auf den Detektor, wobei ein Ereignis, das vom Detektor empfangen wird, ein Poisson-Strom mit einer mittleren Zählrate ist, die ausgedrückt wird als

wo mi und qi eine Dosis der schwachen Quelle bzw. einen Raumwinkel der schwachen Quelle mit Bezug auf den Detektor kennzeichnen, i einen Index der schwachen Quelle kennzeichnet und n die Anzahl von schwachen Quellen kennzeichnet;

Schritt 1.2: Ausrichten der Impulse mit einem konstanten Koeffizientenidentifikationsverfahren oder einem Vorderflankendiskriminationsverfahren; und

Schritt 1.3: Mitteln der ausgerichteten Impulse;

Schritt S2: Berechnen eines posterioren Wahrscheinlichkeitslogarithmuswertes für einen gegebenen Energiewert auf Basis des Szintillationsimpulsrauschmodells, das Folgendes umfasst:

Schritt 2.1: Laden eines Szintillationsimpulssegments S0 und Berechnen einer Wahrscheinlichkeitsfunktion des gegebenen Energiewertes auf Basis des Impulsrauschmodells, wobei das Szintillationsimpulssegment bei einem Zeitursprung t0, an dem das Szintillationsimpulssegment einen Schwellwert Vt nach oben überschreitet, beginnt und das Szintillationsimpulssegment zu einem Zeitpunkt .t0 + Δt endet, wobei Δt größer ist als das Zweifache einer Zeitkonstante einer fallenden Flanke eines Szintillationskristalls; und

Schritt 2.2: Berechnen von Logarithmen für alle Zeitpunkte und Addieren der Logarithmen, um einen Wert einer Funktion zu erhalten, die eine selbe Monotonizität aufweist wie eine posteriore Wahrscheinlichkeitsfunktion; und

Schritt S3: wiederholtes Durchführen des Schritts S2, um einen Energiewert zu erhalten, der eine maximale posteriore Wahrscheinlichkeitsbedingung erfüllt, das Folgendes umfasst:

Schritt 3.1: Berechnen von posterioren Wahrscheinlichkeiten für verschiedene vorläufige Energiewerte durch wiederholtes Durchführen des Schritts S2 als ein Modul;

Schritt 3.2: lineares Suchen nach dem Energiewert, der die maximale posteriore Wahrscheinlichkeitsbedingung erfüllt; und
dadurch gekennzeichnet, dass es Folgendes umfasst:
Schritt 3.3: nachdem der Energiewert erhalten wurde, Korrigieren des Zeitursprungs für Schritt 2.1 auf Basis des Energiewertes und Wiederholen von Schritt 3.1 und Schritt 3.2.


 
2. Verfahren zum Wiederherstellen von Szintillationsimpulsinformationen nach Anspruch 1, wobei die Szintillationsimpulsdatenbank in Schritt S1 2000 oder mehr Abtastwerte umfasst.
 
3. Verfahren zum Wiederherstellen von Szintillationsimpulsinformationen nach Anspruch 1, wobei das Szintillationsimpulsrauschmodell in Schritt S1 Eigenschaften einer Szintillationsimpulsform, einer Szintillationsimpulshöhe und eines Szintillationsimpulsrauschpegels umfasst.
 
4. Verfahren zum Wiederherstellen von Szintillationsimpulsinformationen nach Anspruch 1, wobei in Schritt S1 Energiewerte von Szintillationsimpulsen, die einem selben Energiewert an einem selben Zeitpunkt entsprechen, dispergiert verteilt werden, nachdem die Impulse ausgerichtet wurden, die Verteilung als ein Rauschmodell der Impulse definiert ist und ein Parameter des Rauschmodells verwendet wird, um einen posterioren Wahrscheinlichkeitswert für einen gegebenen Energiewert zu berechnen.
 
5. Verfahren zum Wiederherstellen von Szintillationsimpulsinformationen nach Anspruch 1, wobei Schritt S1 ferner Folgendes umfasst:

Einstellen von einigen Konformitätsimpulsen, die innerhalb eines Energiebereichs ausgewählt werden, als einige einzelne Ereignisse, und

Vorabspeichern und offline Analysieren von solchen digitalisierten elektrischen Impulsen, wobei ein mittleres Impulssignal durch Ausrichten der einigen Impulse erhalten wird und Energiespektren der einzelnen Ereignisse durch Addieren der digitalisierten elektrischen Impulse durch Übernehmen des mittleren Impulses als eine Systemreaktion erhalten werden.


 
6. System zum Wiederherstellen von Szintillationsimpulsinformationen, das Folgendes umfasst:

ein Rauschmodellmodul (100), das dazu ausgelegt ist, vorab erfasste Daten mit einer niedrigen Zählrate zu trainieren, um ein Rauschmodell eines Szintillationsdetektorsystems zu erfassen, wobei das Rauschmodell einen mittleren Impuls, eine Rauschvarianz und eine Rauschverteilungsschiefe repräsentiert und das Rauschmodell dem posterioren Wahrscheinlichkeitsmodul bereitgestellt wird;

ein posteriores Wahrscheinlichkeitsmodul (200), das dazu ausgelegt ist, eine posteriore Wahrscheinlichkeit zu berechnen, um einen posterioren Wahrscheinlichkeitslogarithmuswert für einen gegebenen Energiewert zu erhalten; und

ein Energiewertsuchmodul (300), das dazu ausgelegt ist, nach einem Energiewert zu suchen, der eine maximale posteriore Wahrscheinlichkeitsbedingung erfüllt, wobei das Energiewertsuchmodul Folgendes umfasst:

ein vorläufiges Energieberechnungsmodul (310), das dazu ausgelegt ist, durch Aufrufen des posterioren Wahrscheinlichkeitsmoduls einen posterioren Wahrscheinlichkeitslogarithmuswert für einen vorläufigen Energiewert zu berechnen;

ein lineares Suchmodul (320), das dazu ausgelegt ist, einen Energiewert, der eine maximale posteriore Wahrscheinlichkeitsbedingung erfüllt, mit einem linearen Suchalgorithmus zu berechnen; und dadurch gekennzeichnet, dass es Folgendes umfasst:
ein Zeitursprungskorrekturmodul (330), das dazu ausgelegt ist, einen Zeitursprung für einen Abtastwert zu korrigieren und das vorläufige Energieberechnungsmodul und das lineare Suchmodul wiederholt aufzurufen.


 
7. System zum Wiederherstellen von Szintillationsimpulsinformationen nach Anspruch 6, wobei das Rauschmodellmodul Folgendes umfasst:

ein mittleres Impulsmodul, das dazu ausgelegt ist, einen mittleren Impuls von einzelnen Ereignissen zu berechnen;

ein Impulsrauschmodul, das dazu ausgelegt ist, eine Rauschvarianz von Impulsen der einzelnen Ereignisse zu berechnen; und

ein Wahrscheinlichkeitsdichteverteilungsmodul, das dazu ausgelegt ist, eine Wahrscheinlichkeitsdichteverteilungsfunktion von Rauschen zu schätzen.


 
8. System zum Wiederherstellen von Szintillationsimpulsinformationen nach Anspruch 6, wobei das posteriore Wahrscheinlichkeitsmodul Folgendes umfasst:

ein Wahrscheinlichkeitsfunktionsverteilungsmodul, das dazu ausgelegt ist, auf Basis des gegebenen Energiewertes und des Impulsrauschmodells, das vom Rauschmodellmodul eingegeben wird, einen Wahrscheinlichkeitsfunktionswert jedes Punktes zu berechnen; und

ein Zeitpunktproduktmodul, das dazu ausgelegt ist, einen posterioren Wahrscheinlichkeitslogarithmuswert zu berechnen.


 


Revendications

1. Procédé permettant de recueillir des informations d'impulsions de scintillation, comprenant :

étape S1 : acquérir une base de données d'impulsions de scintillation d'événements isolés de conformité non empilés pour un faible nombre de coups, et construire un modèle de bruit d'impulsion de scintillation pour la base de données d'impulsions de scintillation des événements isolés, dans lequel une impulsion moyenne est calculée pour la base de données d'impulsions de scintillation des événements isolés de conformité non empilés, et une information de forme d'impulsion de scintillation est donnée par l'impulsion moyenne, dans lequel l'étape de calcul de l'impulsion moyenne comprend :

étape 1.1 : réduire une dose de rayonnement d'une source de rayonnement pour obtenir une source faible, et réduire le nombre de photons de haute énergie capturés par un détecteur au moyen de la source faible ou par ajustement d'un angle solide de la source faible par rapport au détecteur, dans lequel l'événement reçu par le détecteur est un flux poissonien ayant un taux de comptage moyen exprimé de la manière suivante :

mi et qi désignent respectivement une dose de la source faible et un angle solide de la source faible par rapport au détecteur, i désigne un indice de la source faible, et n désigne le nombre de sources faibles ;

étape 1.2 : aligner les impulsions avec une méthode d'identification de coefficient constant ou une méthode de discrimination de front montant ; et

étape 1.3 : calculer la moyenne des impulsions alignées ;

étape S2 : calculer une valeur de logarithme de probabilité a posteriori d'une valeur d'énergie donnée, sur la base du modèle de bruit d'impulsion de scintillation, comprenant :

étape 2.1 : charger un segment d'impulsion de scintillation S0, et calculer une fonction de vraisemblance de la valeur d'énergie donnée, sur la base du modèle de bruit d'impulsion, dans lequel le segment d'impulsion de scintillation commence à une origine de temps t0 à laquelle le segment d'impulsion de scintillation va vers le haut d'un seuil Vt, et le segment d'impulsion de scintillation se termine à un point temporel .t0 + Δt, Δt étant supérieur à deux fois une constante de temps d'un flanc descendant d'un cristal de scintillation ; et

étape 2.2 : calculer des logarithmes pour tous les points temporels et additionner les logarithmes pour obtenir la valeur d'une fonction ayant la même monotonie qu'une fonction de probabilité a posteriori ; et

étape S3 : réaliser l'étape S2 de façon répétée pour obtenir une valeur d'énergie satisfaisant une condition de probabilité a posteriori maximale, comprenant :

étape 3.1 : calculer les probabilités a posteriori de différentes valeurs d'énergie provisoire en réalisant de façon répétée l'étape S2 en tant que module ;

étape 3.2 : rechercher linéairement la valeur d'énergie satisfaisant la condition de probabilité a posteriori maximale ; et
caractérisé en ce qu'il comprend :
étape 3.3 : après que la valeur d'énergie a été obtenue, corriger l'origine de temps pour l'étape 2.1, sur la base de la valeur d'énergie et répéter l'étape 3.1 et l'étape 3.2.


 
2. Procédé permettant de recueillir des informations d'impulsions de scintillation, selon la revendication 1, dans lequel la base de données d'impulsions de scintillation à l'étape S1 comprend 2000 échantillons ou plus.
 
3. Procédé permettant de recueillir des informations d'impulsions de scintillation, selon la revendication 1, dans lequel le modèle de bruit d'impulsion de scintillation à l'étape S1 comprend des caractéristiques de forme d'impulsion de scintillation, de hauteur d'impulsion de scintillation et de niveau de bruit d'impulsion de scintillation.
 
4. Procédé permettant de recueillir des informations d'impulsions de scintillation, selon la revendication 1, dans lequel, à l'étape S1, après que les impulsions ont été alignées, les valeurs d'énergie d'impulsions de scintillation correspondant à une même valeur d'énergie en un même point temporel sont distribuées de manière dispersée, la distribution est définie comme un modèle de bruit des impulsions, et un paramètre du modèle de bruit est utilisé pour calculer une valeur de probabilité a posteriori pour une valeur d'énergie donnée.
 
5. Procédé permettant de recueillir des informations d'impulsions de scintillation, selon la revendication 1, dans lequel l'étape S1 comprend en outre les étapes suivantes :

définir un couple d'impulsions de conformité, choisies à l'intérieur d'une gamme d'énergie, en tant que couple d'événements isolés, et

préstocker, puis analyser hors ligne ces impulsions électriques numérisées, dans lequel un signal impulsionnel moyen est obtenu par alignement du couple d'impulsions, et des spectres d'énergie des événements isolés sont obtenus par addition des impulsions électriques numérisées, l'impulsion moyenne étant prise comme réponse de système.


 
6. Système permettant de recueillir des informations d'impulsions de scintillation, comprenant :

un module de modèle de bruit (100)

conçu pour conditionner des données préalablement acquises, ayant un faible taux de comptage, afin d'acquérir un modèle de bruit de système détecteur de scintillation, dans lequel le modèle de bruit représente une impulsion moyenne, une variance de bruit et une asymétrie de distribution du bruit, et le modèle de bruit est fourni au module de probabilité a posteriori ;

un module de probabilité a posteriori (200) conçu pour calculer une probabilité a posteriori afin d'obtenir une valeur de logarithme de probabilité a posteriori pour une valeur d'énergie donnée ; et

un module de recherche de valeur d'énergie (300) conçu pour rechercher une valeur d'énergie satisfaisant une condition de probabilité a posteriori maximale, dans lequel le module de recherche de valeur d'énergie comprend :

un module de calcul d'énergie provisoire (310) conçu pour calculer une valeur de logarithme de probabilité a posteriori pour une valeur d'énergie provisoire en appelant le module de probabilité a posteriori ;

un module de recherche linéaire (320) conçu pour calculer une valeur d'énergie satisfaisant une condition de probabilité a posteriori maximale à l'aide d'un algorithme de recherche linéaire ; et caractérisé en ce qu'il comprend :
un module de correction d'origine de temps (330) conçu pour corriger une origine de temps d'un échantillon et appeler de façon répétée le module de calcul d'énergie provisoire et le module de recherche linéaire.


 
7. Système permettant de recueillir des informations d'impulsions de scintillation, selon la revendication 6, dans lequel le module de modèle de bruit comprend :

un module d'impulsion moyenne conçu pour calculer une impulsion moyenne d'événements isolés ;

un module de bruit d'impulsions conçu pour calculer une variance de bruit d'impulsions des événements isolés ; et

un module de distribution de densité de probabilité conçu pour estimer une fonction de distribution de densité de probabilité du bruit.


 
8. Système permettant de recueillir des informations d'impulsions de scintillation, selon la revendication 6, dans lequel le module de probabilité a posteriori comprend :

un module de distribution de fonction de vraisemblance conçu pour calculer une valeur de fonction de vraisemblance de chaque point, sur la base de la valeur d'énergie donnée et du modèle de bruit d'impulsion introduit par le module de modèle de bruit ; et

un module de produit de point temporel conçu pour calculer une valeur de logarithme de probabilité a posteriori.


 




Drawing
































Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Non-patent literature cited in the description