METHOD FOR ENERGY CALIBRATION OF A SPECTROMETRY DETECTOR

Description

DOMAINE TECHNIQUE

[0001] The technical field of the invention is spectrometry applied to the detection of radiation. ART ANTERIEUR

[0002] Ionizing radiation detection devices, based on gaseous, semiconductor, or scintillator detector materials, generate electrical pulses formed by interactions of radiation within the detector material. The amplitude of each pulse depends on the energy deposited by the radiation during each interaction. These devices are typically coupled to spectrometric measurement circuits. Such devices produce a spectrum, which is a histogram of the amplitudes detected during an exposure period. The spectrum is established using different amplitude channels, usually several hundred or several thousand channels. Each channel corresponds to a narrow amplitude band. Each channel is assigned a value corresponding to the number of pulses detected during the exposure period whose amplitude falls within the amplitude band corresponding to that channel.

[0003] Spectrometric measurement systems are now widely industrialized. Their applications are broad, including measurements of waste, nuclear equipment or facilities, and radiological environmental monitoring. Software allows for pulse processing parameterization, as well as automated control and analysis of measurements.

[0004] However, some steps are delicate and still require considerable manual adjustment. One example is energy calibration, which establishes a relationship between the amplitude of a pulse and the energy released by the interaction that generated the pulse. The amplitude-energy relationship depends on the detector used, as well as the spectrometric measurement circuit parameters, such as amplification gain, pulse shaping parameters, and the number of channels. Furthermore, even with the same device, without adjusting the settings, the amplitude-energy relationship can drift, necessitating periodic calibration.

[0005] Energy calibration is usually performed by exposing a detector to a calibration object with known emission energies. The calibration object may include one or more isotopes with known emission energies. Calibration involves acquiring a spectrum of photons emitted by the isotopes and then identifying the main peaks present in the spectrum. Some of the detected peaks correspond to a known emission energy. Each detected peak extends on either side of a central channel. Energy calibration allows the determination of different channel (i.e., the central channel of a peak) - emission energy pairs. From these different pairs, an analytical function, called the calibration function, is determined, relating the rank of each channel to an energy. Document US2020 / 379133 A1 describes an energy calibration of a spectrum based on the use of a Deep Learning artificial neural network.In particular, the neural network is used to determine the peak positions of emission from point sources with known emission energies. The next step involves adjusting a calibration function to obtain a correspondence between each energy and the position of each peak.

[0006] Energy calibration is a frequent operation that must be performed accurately to avoid distorting the interpretation of measurements. The invention described below facilitates the automation and repeatability of energy calibration. It can also maximize the number of peaks considered when determining the calibration function. EXPOSE DE L'INVENTION

[0007] A first object of the invention is a method for processing a calibration spectrum, according to claim 1, formed by a spectrometric measuring device, the device comprising: a detector, configured to detect particles, and to form, at each detection, a pulse whose amplitude depends on an energy of a particle that has interacted in the detector; a spectrometric measurement circuit, configured to form a spectrum, the spectrum corresponding to a number of particles detected in channels, each channel being assigned a rank, each rank corresponding to a pulse amplitude; The process involves the following steps: a) positioning the device facing a calibration object, the calibration object emitting particles of different known emission energies towards the detector; b) detection of a portion of the particles by the detector and formation of a calibration spectrum of the detected particles, the spectrum comprising different peaks, each peak corresponding to a known energy; c) assignment of a channel rank to each detected peak; d) from the channel ranks assigned to different peaks in step c) and the emission energies, determination of a calibration function, the calibration function determining an energy from the rank of each channel, the calibration function being a bijective function; the process being characterized in that step d) comprises the following substeps: di) taking into account an analytical model of the calibration function, the analytical model being parameterized by a set of parameters; d-ii) for different values ​​of each parameter: application of the calibration function to the rank of each channel assigned to each detected peak, so as to obtain, for each rank, an energy determined by the calibration function; or application of the reciprocal function of the calibration function to each emission energy, so as to obtain, for each emission energy, a channel determined by the reciprocal function; d-iii) determination of the value of each parameter for which: the energies determined during d-ii) are closest to the emission energies of the calibration object; the channels determined during d-ii) are closest to the channels assigned to each detected peak.

[0008] According to the invention as defined in claim 1, substep d-iii) comprises: calculation of a likelihood function of the value of each parameter; estimation of the value of each parameter maximizing the likelihood function.

[0009] Step c) may involve determining the width of each peak. The likelihood function can then take into account the width of each peak.

[0010] Substep d-iii) may include: taking into account different pairs, each pair having an emission energy and a channel assigned to a peak, following step c); determination of the value of each parameter for which the calibration function allows to link several pairs, each determined value forming an initial value of each parameter; for each initial value, definition of a search domain extending around said initial value, so that the value of each parameter maximizing the likelihood function is estimated in the search domains defined around the initial values ​​of said parameter.

[0011] Substep d-iii) is preferably implemented by an optimization algorithm.

[0012] In one possibility, step c) involves selecting peaks from the calibration spectrum based on a peak selection criterion. The selection criterion could be the number of photons detected in each peak or a signal-to-noise ratio determined for each peak.

[0013] Substep d-iii) may include taking into account a priori on the value of at least one parameter to determine the value of said parameter.

[0014] The particles can be chosen from: photons, neutrons, charged particles.

[0015] A second object of the invention is a device for acquiring a spectrum of particles emitted by an object, the device comprising: a detector, configured to detect particles, and to form, at each detection, a pulse whose amplitude depends on an energy released by the particle that interacted in the detector; a spectrometric measurement circuit, configured to form a spectrum, the spectrum corresponding to a distribution of the amplitudes of the pulses detected by the detector; a processing unit, programmed to implement step d) of a process according to the first object of the invention from the spectrum.

[0016] A third object of the invention is a medium, intended to be connected to a computer, containing instructions for carrying out step (d) of a method according to the first object of the invention, based on a spectrum representative of the energy of detected particles. The medium can be integrated into the computer or connected to the computer by a wired or wireless link.

[0017] The invention will be better understood by reading the explanation of the examples of embodiment presented, in the continuation of the description, in connection with the figures listed below. FIGURES

[0018] There figure 1 represents an example of a device enabling the implementation of the invention. The figure 2A represents an example of a spectrum. The figure 2B is a detail of the figure 2A . There figure 3 It shows a peak in a spectrum along with the application of a second-order linear differential operator to the spectrum. figure 4 shows the main steps of an embodiment of a process according to the invention. figure 5 illustrates correspondences between emission energies and central channels of detection peaks. The figure 6 illustrates the correspondences between central channels of detection peaks (x-axis) and emission energies (y-axis). figures 7A And 7Bshow a calibration spectrum acquired with a germanium detector. The positions of channels assigned to different detected peaks are also shown. figure 7C shows the value of a likelihood function for a parameter of the calibration function for different values ​​of that parameter. figure 7C corresponds to the spectrum represented on the figures 7A And 7B . There figure 7D is a detail of the figure 7C . There figure 7E shows the value of a likelihood function of another parameter of the calibration function for different values ​​of that parameter. figure 7E corresponds to the spectrum represented on the figures 7A And 7B . There figure 7F is a detail of the figure 7E . There figure 7G illustrates the correspondences between channels assigned to detection peaks (x-axis) of the spectrum represented on the figure 7A and emission energies (y-axis). The figure 8A This shows a calibration spectrum acquired with a LaBr3 (Lanthan Bromide) detector. Also shown are the positions of channels assigned to different detected peaks and a range of uncertainty relative to the position. figure 8B shows the value of a likelihood function for a parameter of the calibration function for different values ​​of that parameter. figure 8B corresponds to the spectrum represented on the figure 8A . There figure 8C is a detail of the figure 8B . There figure 8D shows the value of a likelihood function of another parameter of the calibration function for different values ​​of that parameter. figure 8D corresponds to the spectrum represented on the figure 8A . There figure 8E is a detail of the figure 8D . There figure 8F illustrates the correspondences between channels assigned to detection peaks (x-axis) of the spectrum represented on the figure 8A and emission energies (ordinate axis). EXPOSE DE MODES DE REALISATION PARTICULIERS

[0019] We have represented, on the figure 1 A device 1 enabling the implementation of the invention. The device is a measurement system, comprising a detector 10, capable of interacting with radiation 5 emitted by an object 2. The object 2 is here a nuclear waste, which may contain different radioactive isotopes 2j. Generally, the radioactive isotopes likely to be present in a measured object are known beforehand. In particular, a list can be compiled containing the radioactive isotopes potentially present in the analyzed object. The subscript j designates each radioactive isotope.

[0020] In the example shown, detector 10 is configured to detect ionizing photon radiation. Ionizing photon radiation is understood to mean X-ray or gamma-ray radiation, consisting of photons with energies, for example, between 1 keV and 2 MeV. The method according to the invention is applicable to the detection of other particles, for example neutrons, charged particles (e.g., alpha or beta radiation), or photons.

[0021] In the example shown, the detector consists of a semiconductor material, such as germanium (Ge), but it could also be a semiconductor material commonly used for detecting ionizing photons, for example, silicon (Si), chlorinated tetrahydrogen (CdTe), or chlorinated zinc (CdZnTe). The photons forming the incident radiation interact within the detector material. The detector material is subjected to a bias voltage V. Each interaction generates charge carriers, which are collected by an electrode, usually an anode.

[0022] Other types of detectors, for example scintillators coupled to a photon / charge carrier converter, or a gaseous detector of the ionization chamber type, can be used, provided that they allow the collection of a quantity of charges Q under the effect of an energy E released by the ionizing radiation during an interaction in the detector 10. Among the usual scintillator type detectors, we can mention Nal(TI) or LaBr 3.

[0023] The detector 10 is connected to an electronic circuit 12, configured to generate a pulse whose amplitude depends on, and is preferably proportional to, the amount of charge collected during an interaction. The amount of charge corresponds to the energy deposited by the radiation during the interaction.

[0024] In the example shown, detector 10 is connected to a cryostat 16, containing liquid nitrogen to maintain detector Ge at an operating temperature.

[0025] The electronic circuit 12 is connected to a spectrometry unit 13, located downstream of the electronic circuit, which allows for the collection of all the pulses formed during an acquisition period. Each pulse corresponds to an interaction of the incident radiation with the detection material. The spectrometry circuit then classifies the pulses according to their amplitude. AThis provides a histogram showing the number of detected pulses as a function of their amplitude. This histogram is an amplitude spectrum. It is usually obtained using a multichannel analyzer. Each amplitude is discretized into channels, with each channel assigned an amplitude band. The value of each channel in the spectrum corresponds to a number of pulses whose amplitude lies within the amplitude band assigned to that channel. Each amplitude band corresponds to an energy band, the correspondence being bijective. Thus, each channel is assigned either an energy band or an amplitude band.

[0026] The relationship between amplitude and energy can be established by irradiating the detector with a calibration object emitting radiation of known energy. Specifically, this radiation must exhibit at least one discontinuity, or energy peak, at a known energy level. This operation is commonly referred to as energy calibration. In gamma spectrometry, the detector is exposed to a 152<Eu type calibration source, producing photons with known emission energies. Other isotopic sources can also be used, as described later in one of the implementation examples. It is preferable that the isotope or isotopes be such that photons are emitted at least two different emission energies. The emission energies range from a minimum to a maximum value.It is preferable that the difference between the minimum and maximum values ​​covers the spectral range in which the detector is intended to be used.

[0027] Energy calibration allows a calibration function to be established. f e allowing an analytical relationship to be established between the measured amplitude and the energy. This takes into account the calibration function. f e This allows, through a change of variable, the assignment of channel values ​​to energy values ​​instead of amplitudes. Indeed, amplitudes depend on the detector and the settings made, while energies correspond to fixed physical quantities, independent of the detector.

[0028] The Specter yis a histogram of the amplitudes of each detected pulse, discretized according to energy or amplitude channels. Each channel is assigned an amplitude band which, by applying the calibration function, becomes an energy band. The spectrum y can be expressed as a vector ( y 1, ... y k ... , y n ), Or n corresponds to the total number of channels. Each channel is assigned a rank. k, with 1 ≤ k ≤ n. Each row channel k is delimited by a lower amplitude A k and a greater amplitude A k +1, such that a detected pulse is assigned to the channel of rank k when its amplitude is between A k And A k +1 .

[0029] The lower amplitude A k each channel corresponds to a lower energy e k The upper amplitude A k +1 for each channel corresponds to a higher energy e k +1. The amplitude-energy correspondence is established by the calibration function. f e described in relation to previous art. Thus, e k = f e k − 1 n β And e k + 1 = f e k n β β is a set of parameters of the function f e as described later.

[0030] The size k − 1 / 2 n corresponds to a position of the center of a channel of rank k relative to the number n of channels in the spectrum. This is a normalized rank for each channel, between 1 (k=1) and (k= n Normalization allows us to establish a calibration function that is independent of the number of channels. n, the latter being configurable.

[0031] The device includes a processing unit 14, programmed to implement steps of the algorithm described in connection with the figure 4 The processing unit 14 is connected to the spectrometry unit 13, from which it receives each measured spectrum.

[0032] The objective of a gamma spectrometry measurement is to identify the 2j isotopes present in object 2 and, preferably, to estimate their respective activities. figure 2A represents a spectrum y resulting from a germanium detector exposed to a source containing the 152Eu isotope. The x-axis corresponds to the channels and the y-axis corresponds to the number of pulses whose amplitude corresponds to each channel. figure 2B is a detail of the figure 2A , corresponding to the rectangular area drawn on the figure 2A .

[0033] The Specter yThe spectrum shown comprises several peaks, each peak corresponding to the emission energy of an isotope present in object 2. These peaks constitute the useful information of each spectrum, from which isotopes can be identified and their respective activities quantified. The spectrum also includes a continuum, corresponding to photon scattering within the detector or before reaching it. The continuum corresponds to the portion of the spectrum below and between each peak. The spectrum also contains a background noise component, resulting in statistical fluctuations.

[0034] One of the common steps in spectrum processing is the detection of each peak. On the figure 2B We have represented different peaks. The x-axis corresponds to the amplitude channels, while the y-axis corresponds to the number of pulses detected in each channel. The processing unit can then determine a characteristic channel for each peak, for example, the central channel around which the peak is centered. The central channel is usually the channel with the largest number of detected pulses.

[0035] Each peak is delimited by two channels, constituting the lower bounds respectively inf i and higher sup i of each peak. Subsequently, each peak is assigned to an order i, in order of increasing amplitude. Each peak of order i is identified by the rank channel k i of its central channel, the latter being arranged between the lower terminal inf i and the upper limit sup i of the peak.

[0036] Depending on one possibility, each peak can be delimited manually. The detection and delimitation of each peak can also be performed automatically, for example by implementing a second-order differentiation operator, denoted d 2. The figure 3 represents a part of a spectrum y corresponding to a peak. On the figure 3 We have also represented a second-order differentiation operator applied to the spectrum y , note d 2< (y). The vertical line, which corresponds to a local minimum of d 2< ( y ), allows the channel of rank to be identified k corresponding to the center of the peak. The differentiation operator can be a Savitzky and Golay filter of order 3 and window width 11

[0037] There figure 4 outlines the main steps of a process for determining a calibration function, these steps being described below.

[0038] Etape 100 : acquisition of the calibration spectrum y During this step, the detector is exposed to a calibration object. The calibration object generally contains one or more calibration isotopes with known emission energies.

[0039] Etape 110 : detection of peaks in the calibration spectrum and assignment of a channel to each detected peak. Note k 1 ... . k i ... . k I the channel ranks assigned to each peak detected on the calibration spectrum. I denotes the number of peaks detected on the calibration spectrum y , which correspond respectively to emission energies E' 1 ... E' j ... . E' J calibration isotopes.

[0040] Each peak of the calibration spectrum y extends along a width at mid-height w i An example of full width at half height is shown on the figure 2B .

[0041] The full width at half maximum (FWHM) is also determined during peak detection by conventional means. For example, when peak detection is performed by applying a Savitsky and Golay filter as described in connection with the figure 3 The full width at half maximum (FWHM) of each peak can be estimated by determining the number of channels between two zero crossings of the function d 2< (y) on either side of the local minimum corresponding to the peak. On the figure 3 , a full width at half height w has been represented. The full width at half height can be determined by other common means of spectrometry software, following a manual delimitation of the peaks.

[0042] Etape 120 : consideration of an analytical model of the calibration function f e . During this step, an analytical model of the calibration function is defined, for example a linear function or a polynomial of predetermined degree. The model is parameterized by a set of parameters β .

[0043] We assume that each energy E i corresponding to a detected peak is located at the center of the channel of rank k i which gathers the energy impulses between e i summer i +1. Thus, we assume that E i = e i + e i + 1 2

[0044] Given (1), (2), and (3) E i = f e k i + k i − 1 2 n β = f e k i − 1 2 n β Or : n corresponds to the total number of channels in the spectrum. k i And k i + 1 are the minimum and maximum amplitudes, respectively, of the amplitude channel of rank i corresponding to the energy E i . ki−12 corresponds to a median amplitude of the channel of rank i, which is assumed to correspond to the emission energy E i .

[0045] The calibration function can be linear, in which case the parameter set is β = ( β 0; β 1) and: E i = β 0 + β 1 k i − 1 2 n

[0046] In the case of a polynomial of degree 2, the calibration function is of the type: E i = β 0 + β 1 k i − 1 2 n + β 2 k i − 1 2 n 2 The set of parameters is β = ( β 0 ; β 1; β 2).

[0047] The calibration function f e is necessarily bijective. It is common for channels to be classified according to increasing amplitudes, in which case the calibration function is increasing.

[0048] E i represents the energy corresponding to the median amplitude of the channel of rank k i , by applying the calibration function f e . Thus, the calibration function f e corresponds, to each rank channel k i an energy E i .

[0049] The calibration object emits photons at emission energies E' 1 ... E' j ... E' J .J corresponds to the number of emission energies. Not all peaks, corresponding to emission energies, may be detected by peak detection, for example, if the number of photons in a peak is too low. The ranks k 1 ... . k i ... . k I corresponding to certain detected peaks correspond to certain emission energies E' 1 ... E' j ... E' J .

[0050] There figure 5 represents an example of channels k 1 ... . k i ... . k I ( I = 5) corresponding to each detected peak taken in ascending order ( k i > k i -1), as well as emission energies E' 1 ... E' j ... E' J isotopes, also taken in order ( E' j > E' j +1 ) ascending ( J = 7). The calibration function must maximize the number of occurrences according to which E ′ j = f e k i − 1 2 n β .

[0051] On the figure 5 Each occurrence is represented by an arrow. When two emission energies are close together, they can be associated with a channel of the same rank. On the figure 5 This particular case is illustrated by energies E 1 and E 2. Etape 130 : Establishing a probability density

[0052] For each energy emission E 1 ... E j ... E J corresponds to a channel of rank x 1 ... x j ... x J by applying the function f e -1< , such that x j = nf e − 1 E ′ j , β + 1 2

[0053] The rating f e -1< ( E' j , β ) refers to the fact that f e -1< depends on β .

[0054] As previously mentioned, a detected peak has a certain spectral width. w i , leading to uncertainty about the channel's rank k i . We can consider that for each emission energy E' j the rank k i the channel in which the peak is detected is a variable following a predetermined probability density g, for example Gaussian, such that: g k i x j , σ i = 1 σ i 2 π exp − 1 2 x j − k i σ i 2 Or : x j is the "true" rank of the channel associated with the emission energy peak E j ; k i is the rank assigned to i nth peak detected; σ i = w i ⋅ 2 .

[0055] Starting from (7), we can write: g κ x , σ = ∑ i = 1 I ∑ j = 1 J g k i x j P i j Or : κ is a vector containing all the ranks of the channels corresponding to detected peaks: κ = k 1 ... k i ... k I P ( i,j ) is a probability that a peak of rank k i corresponds to an emission energy E' j ; σ is the vector (σ 1 ... σ i ... σ I ).

[0056] All combinations k i , E' j are considered equally probable. Thus, P i j = 1 IJ

[0057] Etape 140 : establishing a likelihood function and maximization.

[0058] Taking into account (7), (8), and (9), we can write: g κ E ′ , σ , β = ∑ i = 1 I ∑ j = 1 J 1 σ i 2 π IJ exp − 1 2 nf e − 1 E ′ j , β + 1 2 − k i σ i 2

[0059] From this, we deduce a likelihood function of β noted L ( β | κ , ( σ , E ' ) such as : L β κ , σ , E ′ = g κ E ′ , σ , β

[0060] E' corresponds to the emission energies of the radionuclides used for calibration: E' = E' 1 ... E' j ... E' J .

[0061] We can estimate the parameter vector β̂ which maximizes the likelihood function: β ^ = argmax β L β E ′ , κ , σ β ^ = argmax β ∑ i = 1 I ∑ j = 1 J 1 σ i 2 π IJ exp − 1 2 nf e − 1 E ′ j , β + 1 2 − k i σ i 2

[0062] The vector β̂ is the one for which the number of matches between E' And κ is maximized. This amounts to obtaining a maximum number of occurrences according to which E' j = f e ( k i , β̂ ).

[0063] The vector β̂ The resulting data allows us to define the calibration function f e .

[0064] There figure 6 illustrates a situation in which the function f e is linear. The function f e is such that: f e k = β 0 + β 1 k − 1 2 n for each channel of rank k.

[0065] Conversely, an energy E k corresponds to a channel of rank k such that: k − 1 2 n = f e − 1 E k = E k − β 0 β 1

[0066] Expression (13) becomes: β ^ = argmax β 0 , β 1 ∑ i = 1 I ∑ j = 1 J 1 σ i 2 π IJ exp − 1 2 σ i 2 n E ′ j − β 0 β 1 + 1 2 − k i 2

[0067] The optimization algorithm, expressed by (16), can advantageously be implemented around values ​​of β 0 and β1 predetermined. This allows the maximization algorithm to be applied in a predetermined search domain around the predetermined initial values ​​of β 0 and β 1. This helps to avoid "optimization traps", corresponding to the determination of local maxima of β 0 and β 1. This also optimizes computation time. The search domain is defined around each pair of values β 0 and β 1.

[0068] More generally, regardless of the type of calibration function, the vector β̂ is advantageously determined by applying an optimization algorithm, allowing the solution (16) to be made, based on a restricted search domain around discrete initial values ​​of the vector β̂ . The search domain corresponds to a continuous interval extending around each initial value.

[0069] The definition of the restricted research area is described below, in relation to the figure 6 .

[0070] Sur la figure 6 , The x-axis corresponds to the normalized rank k n of each channel. The vertical axis corresponds to the emission energies E' j normalized by the highest emission energy E' j , max with E' j,max = E' 4. Emission energy is detected when a channel is assigned to it.

[0071] The canals were represented by black vertical lines. k i having been identified by searching for peaks on the calibration spectrum. As previously stated, the energy channels are affected by an uncertainty w i which is represented by two thin vertical lines on either side of a black vertical line. The normalized emission energies E j ′ E j , max ′ correspond to horizontal lines. These are exact values, without associated uncertainty.

[0072] We have drawn, in grey, straight lines that pass through at least two intersections between a normalized emission energy E j ′ E j , max ′ and a detection channel k i taking into account the associated uncertainty. Each intersection corresponds to a match E j ′ E j , max ′ , k i We took into account the fact that the calibration function is necessarily increasing when the channel ranks are ordered in increasing order with the amplitude. Thus, we only represented straight lines that could correspond to an increasing function ( β 1 > 0. The line shown in black is the line with the maximum number of intersections, in this case ( E 1 ′ , k 1), ( E 3 ′ , k 3), ( E 4 ′ , k 4).

[0073] There figure 6 This illustrates the definition of a restricted search domain for implementing the optimization algorithm described earlier. It involves determining initial values ​​for the calibration function parameters for which the bijective application of the calibration function generates occurrences between E' j (Or E j ′ E j , max ′ ) And f e ( k i ) or, conversely, between k i And f e -1< ( E' j In other words, we form different pairs, each pair comprising an emission energy and the rank of a channel assigned to a detected peak. The optimization algorithm described in (16) is implemented around initial values ​​of the parameter vector. β̂ of the calibration function for which the calibration function is the function verifying E' j = f e ( k i ) for at least two pairs whose respective energies and ranks differ from one pair to the other: in other words, for two pairs ( E' j ; k i ) And ( E' j' ; k i ), E' j ≠ E' j' And k i ≠ k i ,

[0074] From the figure 6 , we can establish different initial values ​​of β 0 and β 1 with: β 1 = n E j ′ − E j ′ ′ k i − k i ′ And β 0 = E ′ j ′ − β 1 E j ′ − 1 n

[0075] According to one variant, steps 130 and 140 are implemented by selecting the emission energies E j ′ the most intense and / or the channels k i for which the peak areas are the largest, or exhibit a signal-to-noise ratio considered high. This optimizes computation time. According to this variant, the number of emission energies and channels can be limited to a few dozen. For example, I = 30 and J= 20. More generally, the process involves taking into account a selection criterion for each peak. The peaks processed in steps 130 and 140 are those that satisfy the selection criterion.

[0076] According to one variant, in step 140, an a priori assumption is taken into account regarding the values ​​of one or more parameters forming the set β For example, a range of variation is defined for one or more parameters. Each parameter, for which a prior condition is taken into account, is required to fall within the predefined range of variation. Taking a prior condition into account prevents the optimization algorithm from generating unrealistic parameter values. The prior condition constitutes a constraint for the implementation of the optimization algorithm, in addition to the condition that the calibration function be monotonic. Essais expérimentaux

[0077] The inventors implemented the calibration process as previously described using two types of detectors: a Ge (Germanium) type detector, as schematically shown in the diagram. figure 1 and a LaBr 3 (Lanthan Bromide) type detector. The calibration results are now reported.

[0078] In an initial series of tests, the germanium detector was calibrated for energy. A standard source of 152 Eu, with an activity of 35.04 kBq, was used. A calibration spectrum was acquired over a period of 3 hours. The source was positioned 40 cm from the detector. The spectrum was acquired using 8192 channels. The calibration function was assumed to be linear.

[0079] THE figures 7A And 7B represent the channels resulting from the peak detection algorithm. We observe that no peak is detected beyond channel 6600. On the figures 7A And 7BThe x-axis represents the channels and the y-axis represents the number of pulses (or counts) detected and assigned to each channel. This is a classic spectral representation.

[0080] Peak detection yielded 179,700 possible matching pairs of detected peak channel to emission energy. Taking into account that the calibration function is increasing, this results in 82,650 possible pairs. This corresponds to J = 20 (number of emission energies considered) and I = 30 (number of peaks detected), the number N couples of possible pairs, corresponding to an increasing calibration function, is equal to: N couples = IJ IJ − I − J + 1 4

[0081] There figure 7C shows the value of the likelihood function L (y-axis) as a function of β 0 (x-axis). The figure 7D is a detail of the figure 7B The value of β0 maximizing the likelihood function is equal to 0.3689 keV.

[0082] There figure 7E shows the value of the likelihood function L (y-axis) as a function of β 1 (x-axis). The figure 7F is a detail of the figure 7E The value of β 1 maximizing the likelihood function is equal to 1907.2475 keV.

[0083] There figure 7G is a representation similar to the figure 6 previously described. The x-axis corresponds to the ranks of the channels of the detected peaks, as well as the associated uncertainty.

[0084] The y-axis corresponds to emission energies of 152 < Eu (keV). The application of the calibration function takes into account the values β 0 and β 1 optimal allows to maximize the number of matches between the channels of the detected peaks and the emission energies.

[0085] In a second series of tests, the LaBr 3 detector underwent energy calibration. Standard sources of 22 < Na, 57 < Co, 60 < Co, and 133 < Ba were used. The predominant emission energies were 276.4 keV, 302.85 keV, 356.01 keV, 383.85 keV, 511 keV, 1173.23 keV, 1274.54 keV, and 1332.49 keV. The spectrum was acquired using 4096 channels. The calibration function was assumed to be linear.

[0086] There figure 8A This represents the channels resulting from the peak detection algorithm. Eight peaks were detected, corresponding to the number of emission energies. Peak detection yielded 1568 points, taking into account that the calibration function is increasing.

[0087] There figure 8B shows the value of the likelihood function L (y-axis) as a function of β 0 (x-axis). The figure 8C is a detail of the figure 8B The value of β0 maximizing the likelihood function is equal to -120.7273 keV.

[0088] There figure 8D shows the value of the likelihood function L (y-axis) as a function of β 1 (x-axis). The figure 8E is a detail of the figure 8D The value of β 1 maximizing the likelihood function is equal to 1607.1911 keV.

[0089] There figure 8F is similar to figures 7F And 6 The x-axis corresponds to the channel ranks of the detected peaks, along with their associated uncertainty. The y-axis corresponds to the emission energies listed previously. The calibration function, taking into account the values, is applied. β 0 and β 1 allows maximizing the number of matches between the channels of the detected peaks and the emission energies.

[0090] The calibration function may not be a linear function. For example, it may be a polynomial of degree 2 or higher.

[0091] The invention has been described in connection with gamma spectrometry. It can be generalized to the radiation spectrometry of other types of photons, or other types of particles, in particular neutrons or charged particles. It can also be implemented in the detection of non-ionizing photons, for example, photons in the infrared, visible, or near-ultraviolet spectral ranges.

[0092] It is understood that the invention can be applied to mass spectrometry. In this case, the spectrometric detector is configured to generate pulses whose amplitude depends on the mass of particles that have interacted within the detector. The calibration function is established by exposing the detector to a calibration object containing particles of different masses, the latter being known. The calibration function establishes a one-to-one relationship between the amplitude of each pulse and the mass of the particles detected by the detector.

Claims

1. A method for processing a calibration spectrum (y) formed by a spectrometric measurement device (1), the device comprising: - a detector (10), configured to detect particles, and to form, on each detection, a pulse whose amplitude depends on an energy of a particle having interacted in the detector; - a spectrometric measurement circuit (12, 13), configured to form a spectrum, the spectrum corresponding to a number of particles detected in channels, each channel being assigned a rank, a pulse amplitude corresponding to each rank; the method comprising the following steps: - a) arrangement of the device facing a calibration object (2), the calibration object emitting, to the detector, particles of different known emission energies (E'j); - b) detection of a part of the particles by the detector and formation of a calibration spectrum (y) of the detected particles, the spectrum comprising different peaks, each peak corresponding to a known energy; - c) detection of the peaks of the calibration spectrum and assignment of a channel rank (ki) to each detected peak; - d) from the ranks of the channels assigned to different peaks in the step c) and the emission energies ( E j ′ ), determination of a calibration function, the calibration function determining an energy from the rank of each channel, the calibration function being a bijective function; the method being characterized in that the step d) comprises the following substeps: - d-i) the taking into account of an analytical model of the calibration function (fe), the analytical model being parameterized by a set of parameters (β); - d-ii) for different values of each parameter: • application of the calibration function to the rank of each channel assigned to each detected peak, so as to obtain, for each rank, an energy determined by the calibration function; • or application of the function that is the reciprocal of the calibration function to each emission energy, so as to obtain, for each emission energy, a channel determined by the reciprocal function; - d-iii) determination of the value of each parameter for which: • the energies determined in d-ii) are the closest to the emission energies of the calibration object; • the channels determined in d-ii) are the closest to the channels assigned to each detected peak. the method being characterized in that substep d-iii) comprises: - calculation of a function of likelihood of the value of each parameter L(β|κ, σ, E'); - estimation of the value of each parameter (β̂) maximizing the likelihood function.

2. The method as claimed in claim 1, wherein - the step c) comprises a determination of a width (wi) of each peak; - the likelihood function takes into account the width of each peak.

3. The method as claimed in either one of claims 1 and 2, wherein the substep d-iii) comprises: - the taking into account of different pairings ((E'j;ki)), each pairing comprising an emission energy and a channel assigned to a peak, following the step c); - determination of the value of each parameter for which the calibration function makes it possible to link several pairings, each determined value forming an initial value of each parameter; - for each initial value, definition of a search area extending around said initial value, such that the value of each parameter maximizing the likelihood function is estimated in the search areas defined around initial values of said parameter.

4. The method as claimed in any one of the preceding claims, wherein the substep d-iii) is implemented by an optimization algorithm.

5. The method as claimed in any one of the preceding claims, wherein the step c) comprises a selection of the peaks of the calibration spectrum as a function of a criterion of selection of the peak.

6. The method as claimed in claim 5, wherein the selection criterion is a number of photons detected in each peak or a signal-to-noise ratio determined for each peak.

7. The method as claimed in any one of the preceding claims, wherein the substep d-iii) comprises a taking into account of an a priori regarding the value of at least one parameter to determine the value of said parameter.

8. The method as claimed in any one of the preceding claims, wherein the particles are chosen from among: photons, neutrons, or charged particles.

9. A device (1) intended to acquire a spectrum of particles emitted by an object, the device comprising: - a detector (10), configured to detect particles, and to form, on each detection, a pulse whose amplitude depends on an energy released by the particle having interacted in the detector; - a spectrometric measurement circuit (12), configured to form a spectrum (y), the spectrum corresponding to a distribution of the amplitudes of the pulses detected by the detector; the device being characterized in that it comprises - a processing unit (14), programmed to implement the step d) of a method as claimed in any one of the preceding claims from the spectrum.

10. A medium, intended to be connected to a computer, comprising instructions for the implementation of the step d) of a method as claimed in any one of claims 1 to 8 from a spectrum representative of the energy of particles detected.