Classifications

• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01T—MEASUREMENT OF NUCLEAR OR X-RADIATION
  • G01T1/00—Measuring X-radiation, gamma radiation, corpuscular radiation, or cosmic radiation
  • G01T1/16—Measuring radiation intensity
  • G01T1/17—Circuit arrangements not adapted to a particular type of detector
  • G01T1/171—Compensation of dead-time counting losses

Definitions

• the invention concerns the field of high flow spectrometry using x-rays and gamma rays.
• gamma probes in radiation protection, multi-energy imaging in the medical field (e.g. bi-energy scanners), in the field of non-destructive testing and in security applications (e.g. detection of explosive materials using multi-energy radiography) .
• One particular industrial application of the invention is the detection of explosives for examining baggage using moving radiographs. But other applications are possible, in particular during measurements of intense X and/or gamma photonic flows using spectrometry, for example in measuring waste or nuclear fuels.
• the spectrometric sensors concerned by the invention are preferably direct conversion sensors, i.e. the incident X photons on the sensor interact with a polarized semiconductor material (CdTe for example) , and create a cloud of electronic charges (typically 10000 electrons for an X photon of 60 keV) .
• a polarized semiconductor material typically 10000 electrons for an X photon of 60 keV
• the pulses are classified in different channels depending on their amplitude, and an energy value is assigned to each channel.
• the distribution by channels of each interaction corresponds to the energy spectrum of the radiation having interacted with the irradiated object, or energy spectrum of the detected radiation.
• the radiation is preferably an X or gamma photon radiation.
• the detector involves the poor separability, or stacking, phenomenon of interactions detected by the detector at very close moments.
• the more intense the incident photon flow at the detector the higher the interaction rate (number of interactions occurring in the detector per unit of time) .
• the counting rate measured by the detector corresponding to the number of interactions detected per unit of time, increases with the interaction rate, and the probability of obtaining a stack also increases.
• the counting rate practically does not increase any more, and can even decrease, due to the saturation of the detector.
• the counting rate measured by the detector is the counting rate measured by the detector previously defined.
• the interaction rate in the detector is substantially equal to the counting rate measured by the detector, the latter corresponding to the number of events (or number of counts) appearing in the spectrum per unit of time.
• the measured counting rate then no longer corresponds to the flow to which the detector is subj ected .
• Figure 9 is a spectrum measurement at two different flows, which illustrates the problems posed by the stacking phenomenon.
• Curve I corresponds to an incident flow of 6,082xl0 6 photons/s/detector while curve II corresponds to an incident flow of 4,752xl0 4 photons/s/detector.
• the supplied signal deteriorates due to the stacking phenomenon: if two events are detected with too short a time lapse separating them, the system is not able to tell them apart and provides an erroneous signal depending on the energies of the two photons and the time interval separating them.
• Empirical methods are known: one approach is based on calibrating the stacking phenomenon with radioactive sources having known activity.
• This method is based on knowledge of the duration and energy of each pulse, which is a limitation on said method, in particular at high counting rates.
• the invention concerns a method for correcting a measured spectrum of X radiation (Sp mes ) , including : - determining a stack spectrum (Emp) , which is the part of the measured spectrum (Sp mes ), that corresponds only to the stacks alone,
• the invention concerns in particular a method for correcting a measured spectrum of X radiation (Sp mes ) , according to a number of channels Nc, each channel i corresponding to an energy range between Ei and Ei + ⁇ , including:
• the stack spectrum can itself be calculated from the measured spectrum (Sp mes ) and exposure time (Texpo) and dead time (T dea d) data of the system, which is the minimum duration separating two photons, below which only one of the two photons is detected due to a stacking of those two photons.
• the dead time (T dea d) can be determined by simulation or experimentally.
• a method according to the invention can include the calculation of i t (Nj t >l ) corrected spectrums (Sp cor (n)) from the corrected spectrum of order Spcor (n-i) / or from the measured spectrum if there is no preceding corrected spectrum, by difference between the latter and the stack spectrum (Emp) .
• a method according to the invention can in particular be implemented iteratively.
• a corrected spectrum is determined according to an iterative method including :
• a stack spectrum (Emp) is estimated, as a function of the preceding corrected spectrum, or of the measured spectrum if there is no preceding corrected spectrum, and exposure time ( ⁇ ⁇ ) and dead time (T dead ) data,
• a corrected spectrum is calculated, by the difference between the measured spectrum (Sp mes ) and the estimated stack spectrum (Emp) .
• the invention also concerns a device for correcting a measured X radiation spectrum, including:
• the invention concerns in particular a device for correcting a measured X radiation spectrum, comprising a number of channels Nc, each channel i corresponding to an energy range between Ei and Ei + ⁇ ⁇ , including:
• the stack spectrum can be calculated from the measured spectrum (Sp mes ) and exposure time ( ⁇ ⁇ ) and dead time (T dead ) data of the system, minimal duration, separating two photons, below which only one of the two photons is detected.
• Such a device can include means for determining the dead time by simulation.
• a device preferably includes means for determining a corrected spectrum according to an iterative method including the steps already described above.
• the stacking probability can advantageously be calculated using the formula:
• SP CO r (n-i ) (j) is the value, for the channel j, of the preceding corrected spectrum SP CO r (n-i ) , or of the measured spectrum if there is no preceding corrected spectrum.
• the stack spectrum can be calculated using the formula:
• 5ti, j (k) is determined from the inverse function of the stacking function F _1 E i E , the stacking function associating F E i E , with a pair of photons with energies (Ei, E j ), the energy measured as a function of the shift At between the arrival moments of the two photons.
• the function 5t ⁇ , j (k) determines the size of the temporal deviation interval At separating two interactions of energy Ei and E j , the stacking of which leads to a detected energy value E k .
• the stacking function F E i E can be obtained experimentally or by simulation, for example it is estimated by a decreasing affine function of the energy.
• spectrometry device including:
• the stacks are not limited by a modification of the processing circuit of the signal, but by a processing of the measured energy spectrum.
• 5t ⁇ j (k) is independent of k, and is e ual, re ardless of k, to
• FIG. 1 shows a spectrometry device according to the invention
• - figure 2 shows measurements done on an experimental spectrometry system
• - figure 3 shows a model of a stacking function with two photons
• FIG. 4 shows a dead time measurement by adjusting the formula for the counting rate, for a paralyzable system, to experimental data
• FIG. 7 shows a ratio curve of the stacking probability of two photons over the total stacking probability, for a dead time of 62 ns
• FIGS 8A and 8B are spectrum measurements at different flows, before and after correction according to the invention, respectively.
• - figure 10 presents a spectrometry device, with delay line.
• This device a spectrometry chain 1, includes the following elements:
• this sensor is for example provided with two electrodes, at the terminals of which a signal results from an interaction of a radiation or a photon with the material of the sensor,
• each channel i corresponding to an energy range between Ei and Ei + ⁇ ⁇ , ⁇ ( ⁇ ⁇ 0) then corresponding to the energy width of the channel i.
• processing means for example based on delay line circuits, in particular making it possible to shape the signal, can be provided upstream of the analog digital converter.
• a sample of material 100 is arranged between the source and the detector in order to be characterized.
• the means 12 in particular includes a computer or a microcomputer or a calculator programmed to store and process spectrum data and data to implement a method according to the invention, for example the data ⁇ ⁇ and T dea d- More precisely, a central processing unit
• a processing method according to the invention by calculating a stack spectrum, for example using an iterative method as described below in relation to figure 5, and by calculating or estimating a corrected spectrum (Sp cor ) using the difference between stored measured spectrum data ( Spmes ) and stored stack spectrum data (Emp) .
• Sp cor corrected spectrum
• all or part of the processing method according to the invention could be implemented by the means 10, this means being able to be a FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit) .
• This means 12 can make it possible to control the X radiation source 1, to trigger a radiation emission and perform one or several measurements using the detector 2.
• This electronic means 12 can make it possible to perform a synchronous control of the triggering of the radiation source (s) and the detector (s) .
• This means 12 can also make it possible to conduct an estimate of the dead time T dead and the stacking function F E i E , or from its inverse F -1 , experimentally or by modeling, as explained below.
• an operator can select one or several parameters to carry out these operations.
• the operator can in particular select a number of iterations i t to carry out an iterative process according to the invention, or order the stop of the iterations.
• the iterations can also stop automatically according to a convergence criterion.
• An operator can choose, for example using a pull-down menu, a number of iterations to be performed for processing according to the invention.
• Such a device can also implement a delay line making it possible to shape the pulses in a trapezium, for example as described in EP 2071722.
• This device illustrated in figure 10, primarily includes:
• the resistor 14 designates a polarization resistance associated with the detector 2
• a delay line circuit 22 for measuring energy including a delay line 32, a first gain 34, a subtractor 36 and a second gain 38, connected at the outlet of the preamplification circuit, and
• a current pulse measuring circuit 56 connected at the outlet of the preamplification circuit 20 and making the difference between the output and a derivative of the output signal of the preamplification circuit
• the signal is shaped, then digitized by an analog digital converter (ADC) , then classified in the form of a spectrum by a programmable electronic circuit of the FPGA type.
• ADC analog digital converter
• the passage of the shaped signal (so-called trapezium, or temporal, signal) and digitized via the ADC to the spectrum is done in the FPGA.
• T expo be the exposure time, i.e. the time used or measuring time, or acquisition time, during which the energy spectrum is produced from the incident flow at the detector.
• Spmes is for example one of the spectrums that were commented on above in relation to figure 9.
• such a spectrum includes perturbations that come from photon stacking phenomena .
• Sp mes (E) Sp 0 (E) x (1 - P mean ) +Emp(E)
• the factor (1-P mea n) Spo (E) represents the group of photons that have not undergone stacking.
• Emp(E) is the contribution to the measurement, at energy E, of all of the photons of the spectrum Spo that are stacked (this is the "stack spectrum") .
• Emp(E) depends on the incident spectrum Spo and the behavior of the system formed by the sensor 2 and by the electronic means 4 - 10 with regard to the photons detected at very close times, i.e. in a stacking situation.
• the probability for that same incident photon of not stacking with any photon is equal to the product of the probabilities of the different events, because they are independent events.
• the stacking probability P of an incident photon with at least one other photon is therefore equal to:
• N ⁇ S 0 (E) .
• the stack spectrum Emp(E) corresponds to an estimation of the portion of the measured spectrum Sp mes (E) resulting from stacks. It is obtained by quantifying the energy at which a single event, corresponding to two interactions in a stacking situation, is measured.
• a stacking function is the function that associates this pair with the energy measured as a function of the shift At between the arrival moments of the two photons:
• each of the two photons is indeed measured at its own energy (the higher energy one at Max (El, E2) and the lower energy one at min (Ei, E 2 ) ) .
• an event is measured at a variable energy between Ei+ E 2 .
• i ⁇ j and/or i ⁇ k and/or j ⁇ k For example: i ⁇ j and/or i ⁇ k and/or j ⁇ k.
• this inverse function F _1 EiEj (E k ) represents the temporal shift At between two interactions of respective energies Ei and E j such that these interactions are considered a single interaction with energy E k .
• This function F _1 E i Ej can be chosen to be linear.
• the probability can be estimated that at least one of them is in the time interval between F _1 E i Ej (E k ) and F _1 EiEj (E k+ i) in relation to a photon with energy i:
• P ⁇ , j (k) is the contribution, at energy E k of the stacking function, of the photons with energy Ei stacked with those of energy E .
• Pi, j (k) represents, for each channel (k) corresponding to the energy E k , the probability that an event counted in that channel corresponds to a stack of two interactions, separated in time by a duration shorter than T dead , with respective energies Ei and E .
• Pi, j will be called stacking probability and there will be as many stacking probabilities Pi, j as there are pairs (i, j ) , with l ⁇ i ⁇ Nc and i ⁇ j ⁇ Nc
• the stack spectrum Emp is then estimated by adding up, for each channel k, stacking probabilities previously defined for each pair i and j . This can be done iteratively by a loop over all of the pairs (Ei,Ej) .
• Nc denotes the number of energy channels of the spectrometric sensor
• the dead time T dead can first be obtained by:
• the theoretical counting rate is calculated assuming that it is proportionate to I.
• the reference value is estimated for the lowest flow, for which one considers the stacking phenomenon to be negligible (typically a stacking probability for an incident photon less than 1%) ,
• n is the theoretical counting rate, i.e. the counting rate in the absence of stacks.
• Certain devices for processing pulses and forming a spectrum also make it possible to determine a dead time. This size may also be considered a dead time exploitable by the invention.
• Figure 4 illustrates this principle, the adjustment then being done according to a paralyzable model, preferred by the inventors.
• a set of counting measurements (represented by the dots) is done for different incident flow values, adjusted via the current I of the tube X (in mA) .
• the counting rate formula for a paralyzable system is then adjusted on the experimental data by varying the system's dead time T dead -
• the stacking function one can proceed with an estimation, by simulation, of the output pulse shapes emitted by the semiconductor 2 during absorption of a particle and filtering of said signal by the processing electronics.
• the stacking function is then estimated by simulating the system's response when it is subjected to two pulses corresponding to Ei and E j , and by varying the temporal deviation At between those two pulses.
• the stacking function F E i E (At) is then estimated corresponding to the energy E coming from the stack of two energy pulses Ei and E as a function of the temporal deviation At separating those two interactions.
• the digital processing of the pulses can be neglected. In that case, this simulation makes it possible to estimate the shape of the analog signal, corresponding to the inlet of the block 8 of figure 1. Otherwise, one can proceed experimentally :
• affine model Another method is the so-called “affine model.” Indeed, the measured energy decreasing with the shift At between the detection moments of two interactions, the hypothesis can be made of a stacking function for two photons of the affine type, which linearly connects the measured energy and the temporal shift At.
• the function 5t ⁇ ,j (k) determines the size of the temporal deviation At interval separating two interactions with energy Ei and E j , the stacking of which leads to a detected energy value E k .
• This method makes it possible to deduce, from a spectrum measurement Sp mes , degraded by the stacking phenomenon, a corrected spectrum Sp cor -
• Sp cor will be considered as being established from the measured spectrum Sp mes during the first iteration; but, during subsequent iterations, a corrected spectrum Spcor (n) will be established from the corrected spectrum Sp C or (n-i ) obtained during the preceding iteration (assuming 1 ⁇ n ⁇ i t , where i t is the number of iterations) .
• the spectrum Sp cor can be equal to the spectrum Sp mes during the first iteration and, during successive iterations, Sp cor can be equal to the corrected spectrum during the preceding iteration.
• Sp cor Sp mes ⁇
• This method includes 4 input variables:
• a (digitized) spectrum was obtained by measuring, for example using the device of figure 1, and Sp C or can result from said spectrum.
• T expo The integration time T expo . This is a real number given in ms, and it is a physical data of the system, as a function of the radiation source used.
• the dead time of the sensor - electronics system T dead This is a real number given in ms . It has been explained how it can be measured or estimated.
• Ni t The number of iterations to correct the stacks, therefore done iteratively, Ni t .
• This number is for example chosen by an operator of the data processing system. It is possible for this number not to be predetermined.
• the method will then be stopped by the operator, or according to a convergence criterion, by comparing two spectrums corrected during two successive iterations.
• this method provides a corrected stack spectrum, Sp cor (Nit) , also a vector with dimension N c x l :
• the measured spectrum is identified at the corrected spectrum.
• an iterative correction is then done i t times.
• step S4 correction of the stacks (function 2 ) (step S4) to provide a corrected spectrum Sp CO r-
• step S3 makes it possible to provide an output: Emp (which is a vector), by a calculation also done in one step:
• Em/ ) ££; > .(£)
• Sp C or (n-i ) has already been defined above and 5i , j (k) is the function that was previously presented, and which can be calculated according to one of the methods already explained, and preferably by equation (1) .
• step S4 The correction of the stacks (step S4) is done from the vector Sp mes (or S pc0 r (n-i ) if there has been more than one iteration) of the real number P and the vector Emp determined in S3. It results in an output Spcor(n) ⁇
• Step S4 therefore makes it possible to subtract the stack spectrum from the measured spectrum.
• this factor can be equal to 1- Pmean- If this division has not been done, the shape of the spectrum will be kept, but the integral will not be correct .
• the method stops and the last spectrum Sp cor (Nit) obtained is considered to be the desired corrected spectrum. Otherwise, the process is restarted using the last obtained spectrum Sp CO r (n-i ) as the starting point.
• the calculations are done, in particular of Pi, (k) and Emp(k), as a function of the corrected spectrum obtained during the preceding iteration (except the first iteration where Pi, (k) and Emp(k) are calculated as a function of the measured spectrum, which is then considered the corrected spectrum) .
• this number is between 1 and 100, preferably between 2 and 10.
• the process of estimating the stacking function is based on the hypothesis that the stacks with two photons are mostly compared to the stacks with three or more photons .
• FIG. 6 illustrates the variations of P and P 2 , defined above, as a function of the incident flow, expressed in number of photons per pixel and per second .
• the probability of stacking with two photons (curve II) is an increasing function of the flow for the low values thereof.
• the stacking probability P (curve I) is, however, an increasing function of the flow .
• Figure 7 shows the ratio of the probability of stacking with two photons over the total stacking probability .
• Figures 8A - 8B illustrate the performance of the method for correcting stacks according to the invention, in rapid spectrometry.
• Figure 8B shows the same spectrums after correction (the spectrum I' being the corrected spectrum of spectrum I, etc.) in 3 iterations.
• the complete calibration is relatively simple .
• the determination of the two-photon stacking functions does not require additional calibration; it can be based on a model based solely on knowledge of the dead time.
• the invention therefore proposes a rapid correction method that has the interest of having as input only the data for the spectrum Sp mes as well as the data for ⁇ ⁇ and T dea d-

Landscapes

• Physics & Mathematics (AREA)
• Health & Medical Sciences (AREA)
• Life Sciences & Earth Sciences (AREA)
• General Physics & Mathematics (AREA)
• High Energy & Nuclear Physics (AREA)
• Molecular Biology (AREA)
• Spectroscopy & Molecular Physics (AREA)
• Measurement Of Radiation (AREA)

Priority Applications (4)

Applications Claiming Priority (2)

Publications (1)

Family

ID=42352682

Family Applications (1)

Country Status (6)

Cited By (2)

Families Citing this family (7)

Citations (2)

Family Cites Families (5)

• 2009
  • 2009-11-30 FR FR0958522A patent/FR2953298B1/fr not_active Expired - Fee Related
• 2010
  • 2010-11-30 WO PCT/EP2010/068497 patent/WO2011064381A1/en not_active Ceased
  • 2010-11-30 EP EP10785410.1A patent/EP2507651B1/en active Active
  • 2010-11-30 US US13/512,485 patent/US9689994B2/en active Active
  • 2010-11-30 CN CN201080061692.0A patent/CN102713678B/zh active Active
  • 2010-11-30 JP JP2012541458A patent/JP5914347B2/ja active Active

Patent Citations (2)

Non-Patent Citations (6)

Also Published As

Similar Documents

Legal Events