• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    The invention relates to a method for characterizing a sample, by estimating a plurality of characteristic thicknesses, each associated with a calibration material, comprising the following steps: - acquisition of a spectrum of energy transmitted through this sample (Sech), located in an X and / or gamma spectral band, said transmitted spectrum of the sample; for each of a plurality of calibration spectra (Sbase (Lk; L1)), calculating a likelihood from said calibration spectrum (Sbase (Lk; L1)) and the transmitted spectrum of the sample (Sech) each calibration spectrum (Sbase (Lk; L1)) corresponding to the energy spectrum transmitted through a stack of shims each formed of a known thickness of calibration material; estimating the characteristic thicknesses (L1, L2) associated with the sample according to the maximum likelihood criterion.
    公开号:FR3037401A1
    申请号:FR1555438
    申请日:2015-06-15
    公开日:2016-12-16
    发明作者:Andrea Brambilla；Alexia Gorecki；Alexandra POTOP
    申请人:Commissariat a lEnergie Atomique CEA；Commissariat a lEnergie Atomique et aux Energies Alternatives CEA；
    IPC主号:
    专利说明:

    [0001] 1 CHARACTERIZATION OF A SAMPLE BY DECOMPOSITION BASED ON MATE RIAUX. TECHNICAL FIELD The present invention relates to the field of the characterization of a sample by X spectrometry (wavelength less than 10-8 m) and / or gamma (wavelength less than 1041 m). The invention relates more particularly to a determination of a series of characteristic thicknesses, each associated with a calibration material. We commonly speak of decomposition as a base of materials. STATE OF THE PRIOR ART Various solutions are known in the prior art for characterizing a sample, in particular by X and / or gamma spectrometry. It has been shown in particular that for samples with a rather low average effective atomic number, the attenuation function of the sample L * le) can be described, theoretically, as a linear combination of the respective attenuation coefficients. ilmATi, ilmAT2 of two calibration materials MATI and MAT2. In other words, the sample ec of thickness L produces the same attenuation as a thickness Lh i_ of the material EC MATI plus a thickness Lh2 of the material MAT2: IIMAT1 (E) L. * eich 4limAT2 (E) * Lee _ L * li (E) (1) The materials MAT1 and MAT2 are called calibration because their respective attenuation coefficients define a base of decomposition to characterize the sample. The Leich and Lee thicknesses are said to be characteristic because they characterize the sample in the decomposition base. The attenuation function 3037401 2 is the product of a thickness of a material by its attenuation coefficient. Each attenuation coefficient is a function of energy. The attenuation function is considered on a plurality of energy channels (also called energy bands, or energy intervals).
    [0002] Figure 1 illustrates this property. It is a graph representing the abscissa energy kilo-electron-volt (keV) on a logarithmic scale, and the ordinate the value of the attenuation function on a logarithmic scale (without unit). Curve 101 represents the attenuation function y ech YECH (E) * L of a 10 mm thick polytetrafluoroethylene sample. Curve 102 represents the attenuation function u (El L, mAn. * I of a graphite sample of 1.49 mm thickness) Curve 103 represents the attenuation function --MAT2u (E) * Le2c h of an aluminum sample y of thickness 7.66 mm The sum of the curves 102 and 103 corresponds to the curve 101, so that the sample can be defined by Leic h = 1.49 mm and Lee = 7.66 mm, the calibration materials being graphite and aluminum The different attenuation functions are called linear because they do not have any discontinuity (k-edge). The attenuation of the sample is measured by means of the device 200 as shown in FIG. 2. The device 200 comprises a source of electromagnetic radiation 201, emitting an analysis beam 209 in the X and / or gamma spectral band. also includes a detector 203 adapted to count a number of photons received, for each channel of é The source 201 and the detector 203 together form a spectrometer. The sample 202, of thickness L, is disposed between the source 201 and the detector 203, and is traversed by the analysis beam 209. The analysis beam 209 is attenuated by passing through the sample according to the law of Beer-Lambert: SEcH (E) = S0 (E) exP (- ilEcH * L) (2) 3037401 3 S0 (E) is the energy spectrum measured in the absence of the sample, and SEcH (E) is the energy spectrum measured in the presence of the sample, said transmitted spectrum of the sample. The attenuation function of the sample is then defined by: itEcn * L = -In (SEG / (E)) S0 (E) In theory, we measure the attenuation function of the sample, then the function of attenuation of a thickness L1 of the calibration material MATI (called the first attenuation function) and the attenuation function of a thickness 1.2 of the calibration material MAT2 (called the second attenuation function). Next, we search for the coefficients a and b assigned to the first and second attenuation functions to define the attenuation function of the sample as a linear combination of the first and second attenuation functions. We deduce LIch and Lee, or more exactly the estimates of these lengths, LeIch and Lee. In practice, the measurement of energy spectra is tainted with errors. These errors come in particular from photonic noise (statistical fluctuation of the number of photons interacting in the spectrometer), the width of the energy channels detected by the spectrometer, the electronic noise and other imperfections of the spectrometer. These errors are found in the measured attenuation function of the sample, and in the first and second attenuation functions. These errors affect the linearity of the relationship expressed in equation (1), and preclude the estimation of LeIch and Lee characteristic thicknesses as described above.
    [0003] A known solution is to model the detection chain by a response function of the system, to overcome the errors made by it by reversing the function of the system. A disadvantage of this solution is that it is dependent on the quality of said modeling, a high quality modeling being difficult to obtain. US 8,929,508 provides an analytical formula directly providing an estimate of the characteristic thicknesses / h and Le, assuming that the relationship (1) is linear on each of a plurality of small thickness and spectral ranges. transmitted. Such an assumption, however, does not provide a sufficiently accurate estimate of the characteristic thicknesses, especially since it is based on the assumption that the attenuation by the sample in each energy channel follows a Gaussian law. , while a Poisson law is the most realistic hypothesis. An object of the present invention is to provide a method and a device providing an accurate estimate of the characteristic thicknesses used to characterize a sample in a base of calibration materials.
    [0004] DISCLOSURE OF THE INVENTION This objective is achieved with a method of characterizing a sample, by estimating a plurality of so-called characteristic thicknesses, each associated with a so-called calibration material. The method according to the invention comprises the following steps: acquisition of an energy spectrum, said transmitted spectrum of the sample, said spectrum being defined by a number of photons transmitted through the sample in each channel of a plurality of energy channels located in an X and / or gamma spectral band; for each of a plurality of so-called calibration energy spectra, calculating the value of a likelihood function from said calibration spectrum and the transmitted spectrum of the sample, each calibration spectrum corresponding to the transmitted spectrum of the calibration spectrum; a stack of shims, each shim being formed of a known thickness of calibration material; Determining the estimates of the characteristic thicknesses associated with the sample, from said values of a likelihood function and according to the maximum likelihood criterion. This eliminates the limitations associated with measurement errors existing in the prior art described in the introduction. Characteristic thicknesses of calibration materials characterizing a sample in a calibration base can be estimated directly from measured spectra and without resorting to purely theoretical spectra obtained by modeling an acquisition chain. This eliminates inaccuracies related to the imperfections of such modeling. The estimation of the characteristic thicknesses does not rest on any hypothesis which is particularly simplifying. The measurement error is less, thanks to the use of the series of calibration spectra forming a decomposition basis in so-called calibration basis calibration materials. In addition, as detailed in the following, such a method makes it possible to use the correct model of a behavior of the attenuation in each energy channel (behavior that can be called measurement noise, and modeled by a law Poisson statistics). The use of the most accurate model provides an accurate estimate of the characteristic thicknesses. The determination of the estimates of the characteristic thicknesses associated with the sample advantageously comprises a search for a maximum of the values of the likelihood function, the thicknesses associated with this maximum forming the estimates of the characteristic thicknesses.
    [0005] Preferably, the method comprises at least one step of interpolating the values of the likelihood or interpolation function of the calibration spectra. At least one interpolation step can implement a nonlinear interpolation function.
    [0006] The method according to the invention can comprise in particular the following steps: interpolation of the values of the likelihood function by a likelihood interpolation function dependent on at least one variable, each variable corresponding to the thickness of the a calibration material; and searching for a maximum of the values of said likelihood interpolation function, the thicknesses associated with this maximum forming the estimates of the characteristic thicknesses.
    [0007] In addition or alternatively, the method according to the invention may comprise the following steps: interpolation of the calibration spectra by an interpolation function of the spectra depending on at least one variable, each variable corresponding to the thickness of a calibration material; and searching for a maximum of the values of said interpolation function of the spectra, the thicknesses associated with this maximum forming the estimates of the characteristic thicknesses.
    [0008] According to a particular embodiment, the following steps are implemented: interpolation of the values of the likelihood function by a likelihood interpolation function, said values being associated with known combinations of thicknesses of calibration materials such that For each calibration material, the associated thicknesses are located in a respective first interval; Searching for a maximum of the values of said likelihood interpolation function, the thicknesses associated with this maximum forming approximate values of the characteristic thickness estimates; interpolation of the calibration spectra by an interpolation function of the spectra depending on at least one variable, each variable corresponding to the thickness of a calibration material and taking values in a second respective interval, narrower than the first interval associated with the same calibration material and centered on the approximate value associated with the same calibration material; For each of the values of said spectral interpolation function, calculation of the value of the likelihood function and search for a maximum of said values of the likelihood function, the thicknesses associated with this maximum forming consolidated estimates of the characteristic thicknesses .
    [0009] According to another particular embodiment, which can be combined with the preceding one, the following steps are carried out: interpolation of the calibration spectra by an interpolation function of the spectra, said calibration spectra being associated with combinations of thicknesses known from calibration materials such that for each calibration material, the associated thicknesses are located in a respective first interval; for each of the values of said spectral interpolation function, calculating the value of the likelihood function and searching for a maximum of said values of the likelihood function, the thicknesses associated with this maximum forming approximate values of the estimates of the likelihood function, characteristic thicknesses; interpolating the values of the likelihood function by a likelihood interpolation function dependent on at least one variable, each variable corresponding to the thickness of a calibration material and taking values in a second respective interval, narrower than the first interval associated with the same calibration material and centered on the approximate value associated with the same calibration material; And searching for a maximum of the values of said likelihood interpolation function, the thicknesses associated with this maximum forming consolidated estimates of the characteristic thicknesses.
    [0010] The likelihood function is advantageously determined from a statistical modeling of the transmitted spectrum of the sample, according to a Poisson distribution. The likelihood function calculated from said calibration spectrum and the transmitted spectrum of the sample can be defined by: in (V (Sech, sbase (Li, Lm))) R = C ase Lm) -F.Sfch 111 (Sfase (Li, ..., Lm)) 1 = 1 1 = 1 V is the likelihood function, In is the natural logarithm, = R Ei = lech is the transmitted spectrum of the sample, presenting j = R dry 20 energy channels, 's-base Lm) = Eif = isfase (L1, ..., Lm) energy channels of the combination of M calibration materials each associated with a respective thickness 1,1, C is a constant. 25 is preferably used in the method according to the invention, at least one calibration spectrum (Sbase (Li; L2)) corresponding to the transmitted spectrum of an annex standard, this an additional standard being constituted by a determined thickness of a reference material, and being associated with a combination of known thicknesses of calibration materials such that at least one thickness takes on a negative value.
    [0011] The method according to the invention may comprise a step of manufacturing a calibration base comprising said calibration spectra, comprising: measurements of the transmitted spectra of each of a plurality of wedge stacks, each wedge consisting of a known thickness of calibration material; A measurement of the transmitted spectrum of at least one annex standard consisting of a determined thickness of a reference material, and associated with a combination of known thicknesses of the calibration materials such that at least one thickness takes a negative value ; ranking all the spectra measured in a single data base connecting a spectrum to a combination of thicknesses of the calibration materials. The estimation of the characteristic thicknesses can be followed by a calculation of the average effective atomic number of the sample, based on the estimates of the characteristic thicknesses. The invention also relates to a computer program product arranged to implement a method according to the invention. The invention finally relates to a device for characterizing a sample, comprising an electromagnetic source emitting in an X and / or gamma spectral band, and a detector for measuring a transmitted spectrum of the sample, said spectrum being defined by a number of photons transmitted through the sample in each channel of a plurality of energy channels. According to the invention, the device comprises a processor, arranged to implement the method according to the invention, and a memory receiving the calibration spectra, the memory being connected to the processor. BRIEF DESCRIPTION OF THE DRAWINGS The present invention will be better understood on reading the description of exemplary embodiments given purely by way of indication and in no way limiting, with reference to the appended drawings in which: FIG. 1 illustrates the linear attenuation functions a sample and two calibration materials used in the processes according to the prior art; FIG. 2 schematically illustrates a device for measuring an energy spectrum according to the invention; Figure 3 schematically illustrates a first embodiment of the method according to the invention; FIG. 4 schematically illustrates a device according to the invention; FIG. 5 illustrates calibration spectra according to the invention; FIG. 6 schematically illustrates a second method embodiment of the invention; and - Figure 7 illustrates a variant of the embodiment of Figure 6; And Figure 8 illustrates an enriched calibration base according to the invention. DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS The invention relates to a method for characterizing a sample. The characterization here consists in defining a spectrum of the energy transmitted through the thickness sample L, such as the spectrum of the energy transmitted through a stack of two or more shims, each consisting of a calibration material. different.
    [0012] Each shim made of calibration material has a thickness, called the characteristic thickness, or equivalent thickness, which one seeks to estimate and which characterizes the sample. Calibration materials are materials whose respective attenuation coefficients define a decomposition basis for characterizing any sample. Two or more calibration materials are considered. Thus, it is possible to associate with the sample a linear combination of the transmitted spectra of calibration materials, each associated with a characteristic thickness.
    [0013] It is equivalent to associating with the sample a linear combination of attenuation coefficients of calibration materials, each associated with a characteristic thickness (the energy spectrum S0 (E) in the absence sample is of hold being a constant). The method according to the invention is implemented in a device 400 as represented in FIG. 4. The source 401 and the detector 403 correspond to the source 201 and the detector 203 described with reference to FIG. at the distance traveled in the sample 402, respectively in a shim of calibration material by an analysis beam 409 emitted by the electromagnetic radiation source 401 and received by the detector 403 after passing through the sample 402, respectively the hold (see Figure 4). The sample here refers to any object, including a biological sample such as a biological tissue. The method according to the invention does not impose a condition on its average effective atomic number. Nevertheless, it is advantageously less than 30. During an initial calibration step, a series of spectra of the energy transmitted each through a stack of shims is acquired. Each energy spectrum is defined by a number of photons transmitted through the stack, in each channel of a plurality of energy channels located in an X and / or gamma spectral band. Such an energy spectrum is called the calibration spectrum, or calibration spectrum. Each shim, also called standard or calibration sample, is made of a calibration material, and has a known thickness. This is for example a lamella constituted of said material. The multiple shims of the stack are formed of separate calibration materials. This step is preferably carried out only once, the same calibration spectra being used thereafter to characterize any sample.
    [0014] This step is preferably implemented in the device of FIG. 4, under the same measurement conditions as the transmitted spectrum of the sample as defined below (same spectrometer, same transmission power, same distance between source and detector, same irradiation duration). If necessary, several measurements are averaged on the same stack of shims in order to overcome the photonic noise. In a variant, the calibration spectra are calculated from a numerical modeling of the device of FIG. 4, in particular a numerical modeling of the acquisition chain. In a first step 31, the transmitted spectrum of the sample, Si, is acquired by means of the detector 403. The transmitted spectrum of the sample 20 is the spectrum of the energy transmitted through said sample. It is defined by a number of photons per energy channel, for each channel of a plurality of energy channels. The transmitted spectrum of the sample comprises at least two energy channels, preferably several tens. For example, the energy channels together cover the entire energy band from 10 to 120 keV, and each has a width of 1 keV. The energy channels of the transmitted spectrum of the sample are preferably the same as those of the calibration spectra. For each of the plurality of calibration spectra, the value of a likelihood function corresponding to the likelihood 3037401 13 of the calibration spectrum taking into account the transmitted spectrum of the Sech sample is calculated in step 32. The different energy channels of the same spectrum are treated together, and not independently of each other. Since each calibration spectrum is associated with a known thickness of a calibration material, the likelihood function according to the invention depends on the thicknesses of the calibration materials. For example, two calibration materials MATI and MAT2 are considered. A Shase (Li, L2) calibration spectrum is the energy spectrum transmitted through a shim L1 of MATI material contiguous to a shim L2 of material MAT2. The likelihood function calculated from the spectrum transmitted sample S "h and the calibration spectrum Shase (Li; L2) depends on L1 and L2, and Figure 5 shows a series of calibration spectra. values 0, 1, 2 or 3 (unit of length) L2 can take 4 values 0, 1, 2 or 3 (unit of length) Thus, 4x4 = 16 calibration spectra corresponding to all possible combinations of L1 If L1 can take N values, and L2 can take N2 values, we obtain at the end of step 32 N1xN2 values of the likelihood function. likelihood for determining an estimate of a thickness / h of the MAT material and a thickness of the material MAT2, such that the transmitted spectrum of the sample corresponds to the spectrum transmitted through the juxtaposition of a shim. LIch thickness in MATI material and a Lee shim of MAT2 material. note LIch and Le2ch the estimated values of LeIch and Lee.
    [0015] It is a question of searching, thanks to the calibration base, the calibration spectrum composed of a thickness L1 of the material MATI and a thickness L2 of the material MAT2, which most closely resembles the transmitted spectrum of the sample. The criterion used for the estimation is the maximum likelihood.
    [0016] In the example illustrated in FIG. 3, one searches among the N1xN2 values of the previously calculated likelihood function, which is the maximum value. In other words, the calibration base is searched for the calibration spectrum that most closely resembles the transmitted spectrum of the sample. The thicknesses L1, L2 associated with this maximum value of the likelihood function correspond to the Leic h and Lee estimates of the characteristic thicknesses Leic h and Lee. For example, either a thickness L1 of polyethylene taking the values 0, 1, 2 or 3 (length unit), and a thickness L2 of PVC taking the values 0, 1, 2 or 3 (unit of length), we obtain for a given sample the following values of the likelihood function: L2 0 L1 0 4457 5173 5646 5925 1 5891 6068 6121 6075 2 6093 6044 5914 5721 5887 5705 5470 5192 We then have Leich = 2 and Lee = 1, the maximum value of the likelihood function being 6121.
    [0017] Thus, according to the invention, in order to estimate the characteristic thicknesses, a plurality of energy spectra associated with a plurality of combinations of known thicknesses of calibration materials are used. In the general case with M calibration materials, each calibration material being able to take Uy different thicknesses, there is a calibration base 20 including nym_i_ Uy calibration spectra. The number of possible thicknesses for each material may vary. Each calibration spectrum is denoted sbase a1, L2,..., Lm1) aVeC Li, L2,..., L Ni the thicknesses of the materials MATI, MAT2, ..., MATM.
    [0018] The transmitted spectrum of the sample to be characterized is compared to that of the calibration base using the following likelihood function: V (Li, L2, ... Lm) = Flif = 1p (sjechls jbase - (1 , L2, ... LM)) (4) ... Lm) describes the likelihood that the transmitted spectrum of the sample corresponds to the transmitted spectrum of a stack of thicknesses Li, L2, ... Lm of the M materials of the calibration base. ech is the number of photons (or counts) counted in the channel], in the transmitted spectrum of the sample, Sec *, at R energy channels. The R components of the Sec vector * are independent random variables. 10 p (sjechls fuse ... Lm)) is the probability that the channel j of the transmitted spectrum of the sample corresponds to the channel j of the calibration spectrum associated with the thicknesses Li, L2, ... Lm of the materials there M. Thus, the likelihood function is equal to the product, for each of the channels j, of the probabilities that the channel j of the transmitted spectrum of the sample 15 (measured spectrum) corresponds to the channel j of the calibration spectrum associated with the thicknesses L1, L2, ... Lm. Function P describes a statistical modeling of the rate of transmission through the sample, in each energy channel.
    [0019] According to a particularly advantageous embodiment, it is assumed that the arrival of the photons in each energy channel follows a Poisson statistical law. The choice of a Poisson's law makes it possible to offer the best estimate, this law best modeling the physical reality in the spectrometer. Thus, in each energy channel j, the probability that there are ech photons 25 transmitted through the stack of thicknesses Lki of materials MATi, j = M, during a predetermined period of irradiation, is given by: sech With vi the number of photons transmitted by the sample in the channel j, during a irradiation time T (identical for the transmitted spectrum of the sample and for the irradiation time T (identical for the transmitted spectrum of the sample and for calibration spectra). p (sechl) is the probability of measuring ech strokes for expectation y1.
    [0020] If we assume that the thicknesses of materials 1 to M are the thicknesses Li, L2, ... Lm, we have: vi = lis-jbase (L1,1,2, ... Lm,) u translates the drift the spectrometer between the measurement of the calibration spectra and the measurement of the transmitted spectrum of the sample. We assume that this drift is zero, that is, u = 1, a very realistic hypothesis. The likelihood function is then expressed: (Li, L2, ... Lm) = 1 ech 1 (S, ease (Li, L2, ... Lm) sych = 1-1 exp (-Sfase (Li, L2 , ... Lm)) '(7) sechi 1 = 1 - 15 The characteristic thicknesses of the sample are obtained by looking for the maximum of the likelihood function, for convenience it is simpler to try to maximize the logarithm of the likelihood function, which is faster to compute, in particular: ln (V (L1, L2, Lm)) _.s.jbase (LlL2 ... LM) -FS I ch In jb as e ( Li, L2, ... LM)) (8) j = 1 j = 1 Then, the characteristic thicknesses of the M basic materials are given by: geich Le2c h Lee ^ = ar gmax (In (V (Li, L2 , ... Lm))) (9) (6) Sbase (Li, L2, ... Lm)) 3037401 It can be seen that the invention implements a probabilistic approach, based on likelihood calculations and using as estimation criterion the maximum likelihood. This approach is based on Bayes' theorem, and on the assumption that all thickness combinations are equiprobable. This hypothesis makes it possible to establish that: 6 = argmax (P (Sec * 10)) (10) with Sec * that the transmitted spectrum of the sample, 61 = Li, L2, ... LM, and -0 = 10 Leic h, Le2c h,. . . , Ch -11, -1 the estimated characteristic thicknesses. Since the calibration base has only a finite and discrete number of material thicknesses, the likelihoods for the intermediate thicknesses are interpolated by interpolation, either by interpolating the calibration spectra or by interpolating the likelihood. Interpolation limits the number of calibration spectrums required. Linear or polynomial interpolation models, or any interpolation model best describing the behavior of the spectra or likelihood as a function of the characteristic thicknesses of the calibration materials can be used. It can be an interpolation of the calibration spectra. There is thus a greater number of so-called reference spectra used to calculate the value of a likelihood function corresponding to the likelihood of this reference spectrum taking into account the transmitted spectrum of the sample S ech. The reference spectra refer to both the calibration spectra and the interpolated spectra (values of an interpolation function of said calibration spectra). In particular, it is possible to interpolate the calibration spectra by a interpolation function depending on the thickness of one of the calibration materials. It is thus possible to enrich the calibration base by so-called interpolated calibration spectra. The values of a likelihood function are then calculated from the spectrum of the Sech sample and the spectra of the enriched calibration base. The maximum of the values of the likelihood function is then sought, this maximum being associated with the estimates of the characteristic thicknesses. Preferably, the calibration spectra are interpolated by an interpolation function dependent on at least one variable, each variable corresponding to the thickness of one of the calibration materials. More preferably, the interpolation function depends on several variables, corresponding to the thicknesses of each of the calibration materials. One can for example perform a linear interpolation on the logarithm of energy spectra. That is, a calibration base with two calibration materials. That is to say a thickness / 1 of material MATI between L and Lrl, and a thickness / 2 of material MAT2 between Lq2 and 4 + 1. The spectrum transmitted by a thickness / 1 of MATI material attached to a thickness / 2 of material MAT2 can be estimated by interpolating the spectra of the calibration base. For example, one can calculate the spectra in / 1.4 (respectively in / lel) from the spectra in LI, Lq2 and Lr1, Lq2 (respectively in, Lq2 + and Lr1, Lq2 + 1). For each channel j, the number of photons is given by: in (sfase (il, e2ar)) Lk - = lri (S ase Lv2ar) In ase (l k + 1 La r)) 1 1 (11) - 1 with Lrr = Lq2 or Lv2ar = Lq2 + 1 We can then interpolate the spectrum in / 1.12 from the previously calculated spectra in / 1.4 and / 1.4 + 1. The value of the spectrum at the channel j for a thickness / 1 of MATI material joined to a thickness / 2 of MAT2 material is obtained as follows: ## EQU1 ## Alternatively, a nonlinear interpolation is performed, for example by a Lagrangian polynomial or a cubic Hermite polynomial. interpolation by a interpolation function depending on the thickness of a calibration material makes it possible to obtain a calibration spectrum for each thickness of this calibration material situated in a determined interval.
    [0021] The interpolation of the spectra may also consist simply in increasing a number of available spectra by adding to the initial calibration spectra a finite number of spectra obtained by interpolation. Alternatively or additionally, the values of the likelihood function are interpolated. There is thus a greater number of values among which a maximum is sought. In particular, it is possible to interpolate the values of the likelihood function by an interpolation function dependent on at least one variable, each variable corresponding to the thickness of one of the calibration materials. More preferably, the interpolation function depends on several variables, corresponding to the thicknesses of each of the calibration materials. The maximum of the values of the likelihood function are then sought from among the initial values and the values obtained by interpolation. This maximum is associated with the characteristic thickness estimates. For example, the values of the likelihood function are interpolated using a function of two variables. Each variable corresponds to one of the MATI or MAT2 materials. This function is advantageously nonlinear, for example a second order polynomial function of the type: F (Li, L2) = a + bLi + cL2 + dL1L2 + eL21 + 1122 (13) where a, b, c, d, e, f are least squares-adjusted constants, at the values calculated in step 32. The interpolation by a interpolation function depending on the thickness of a calibration material makes it possible to obtain a value of the likelihood function for each thickness of this calibration material within a given range. The interpolation of the values of the likelihood function may also consist simply of increasing a number of available likelihood function values by adding to the initial values a finite number of interpolated values.
    [0022] The interpolations make it possible to use a reduced number of calibration spectra, while offering a high accuracy of the estimation of the characteristic thicknesses. These two types of interpolations can be combined and / or implemented several times to gradually refine the estimation of the characteristic thicknesses (iterative method). For example, an interpolation of the calibration spectra is performed to obtain a plurality of values of the likelihood function, then these values are interpolated and a maximum of the values obtained after this second interpolation is sought.
    [0023] It is also possible to implement several successive interpolations, each time reducing the range of thicknesses considered and the pitch, as a function of the estimate previously obtained. In particular, successive cycles of search for a maximum of the values of the likelihood function and estimation of the characteristic thicknesses are used. At each cycle, a pitch of the calibration material thicknesses associated with the calibration spectra and / or the values of the likelihood function is reduced. Preferably, the decrease of this step is accompanied by a decrease of a range of thickness considered. The calibration spectra here designate the initial calibration spectra and, if appropriate, spectra obtained by interpolation (s). The values of the likelihood function here designate the values calculated from calibration spectra and, if appropriate, values obtained directly by interpolation (s). The detector 403 is connected to a processor 404, adapted to implement the characterization method according to the invention. The processor 404 is connected to a memory 405 storing the calibration spectra. The processor receives as input a transmitted spectrum of the sample, and outputs the estimates of the characteristic thicknesses of the sample.
    [0024] FIG. 6 illustrates an advantageous example of a method employing two successive cycles of searching for a maximum of the values of the likelihood function and estimation of the characteristic thicknesses. In a first step 61, the transmitted spectrum of the sample is acquired, Sech. The values of a likelihood function from this spectrum S "h and each of the sbase calibration spectra (Li, L2) of an initial calibration base are then calculated (step 62). corresponds to stacks of shims with a thickness L1 of the material MATI and a thickness L2 of the material MAT2, with: - L1 extending over a first interval associated with the material MATI 25 (interval [0; LMAT1max1 for example) and according to a first sampling step associated with the material MATI (not LmATi max) and N1 '- L2 extending over a first interval associated with the material MAT2 (interval [0; LMAT2max1 for example) and according to a first step LmAT2max) of associated sampling MAT2 material (not N2.
    [0025] Thus, N1 * N2 are obtained spectra acquired with thickness combinations ranging from 0 to LMATI max for the material MATI and from 0 to LMAT2 max for the material MAT2.
    [0026] In step 65, the values of the likelihood function are interpolated. Then, the maximum of the values of the likelihood function available after interpolation is searched (step 66). This maximum is associated with the thicknesses 1 and 2 respectively of the material MAT1 and material MAT2, and form approximate values of the characteristic thicknesses. In step 67, for each calibration material, a second respective interval is determined, narrower than the first interval associated with the same calibration material, and centered on the corresponding approximate value. The second intervals are obtained [L'11; The 12] and [Un; The 22].
    [0027] The spectra of the initial calibration base are then interpolated on these second intervals [Il; 12] and [21; 22] (step 68). An enriched calibration base is obtained, comprising the Sgbase (Li; L2) spectra. The enriched calibration base corresponds to: a thickness L1 of the MATI material extending over the second gap [Un; 12] (narrower than the [O; LMATI maxi] interval and at a respective second sampling interval lower than the first sampling step associated with the same calibration material (not less than LMAT1 MaX) and N1 H - a thickness L2 of MAT2 material extending over the second gap [L'21; 22] (narrower than the interval [O; LMAT2 max]) respective sampling steps less than the first sampling step associated with the same calibration material (not lower than LMAT2 max) and according to a second Ni.
    [0028] The step 62 of calculating the values of a likelihood function is then reiterated, this time from the Sech spectrum and from each of the spectra of the enriched calibration base Sgbase (Li; L2). Finally, the maximum value thus calculated is sought (step 69). This maximum is associated with thicknesses Leic h and Le2c h respectively of the material MATI and material MAT2, and form consolidated estimates of the characteristic thicknesses. According to a variant represented in FIG. 7, interpolation of the calibration spectra of the initial calibration base Sbase (Li; L2) defined above (step 78) is carried out in first interpolation. An enriched calibration base S''b "e (Li; L2) corresponding to the initial N1 * N2 spectra is obtained, plus values obtained by interpolation.The enriched calibration base is associated, for each calibration material, with a first one. Then, the transmitted spectrum of the sample, Sech (step 71) is acquired, and the values of a likelihood function are calculated from this Sech spectrum and each of the spectra of the base. for enriched calibration (step 72) The maximum of the likelihood function values thus obtained are searched for (step 73) The thicknesses associated with this maximum form approximate values L (L) and (2) L2 of the estimates. characteristic thicknesses.
    [0029] Next, second intervals [L (122; L (122)] (2) (2) and [L21; L22], each associated with one of the calibration materials, are determined at step 77. Each second interval is The second interval is narrower than the first thickness range as defined above, associated with the same calibration material and the enriched calibration base. then performs an interpolation of the values of the likelihood function (step 75) and then searches for the maximum of the values of the likelihood function (new iteration of step 73), this time among the values of the function of After the second interpolation, this maximum is associated with the thicknesses LIch and Le respectively of the MAT material and MAT2 material, and form consolidated estimates of the characteristic thicknesses. As mentioned above, the calibration spectra can be obtained experimentally, using stacks of shims of calibration material of known respective thicknesses, or numerically simulated. This gives a first calibration basis.
    [0030] Each calibration material has an effective atomic number. The maximum and minimum effective atomic numbers of the calibration materials considered together define an effective atomic number interval. If the average effective atomic number of the sample is outside this range, at least one of the associated characteristic thicknesses may be negative. For example, if the material MATI is polyethylene (Zeff = 5.53), the material MAT2 is PVC (Zeff = 4.23), a sample in iron (Zeff = 26), in chromium (Zeff = 24), or in chlorine (Zeff = 17), will be characterized by a characteristic thickness of negative polyethylene. We are then in an area not covered by the first calibration base.
    [0031] A common solution is to extrapolate the points of the first calibration base. We propose here a more reliable solution. It is a question of enriching the first calibration base by means of measurements or simulations involving an appendix of determined thickness consisting of a reference material, distinct from the calibration materials MATI and MAT2. The reference material has an effective atomic number outside the range described above. The characteristic thicknesses of this annex standard are known in the sizing material base, and at least one of these characteristic thicknesses is negative. In other words, the spectrum of energy transmitted through the annex standard is equal to or the spectrum of energy transmitted through a virtual stack of shims of calibration material, one shim or less having a negative thickness (hence the term "virtual"). The first calibration base is thus enriched by points associated with negative characteristic thicknesses. The characteristic thicknesses associated with said annex standard can be determined using a method according to the prior art as described in the introduction. Thus, each calibration spectrum corresponds to the spectrum of the energy transmitted through a stack of wedges each formed of a known thickness of calibration material, if necessary a wedge having a negative thickness (virtual wedge). FIG. 8 shows diagrammatically the points of a calibration base thus enriched. A first series of dots corresponds to combinations of polyethylene and PVC, the thicknesses of polyethylene and PVC taking integer values between 0 and 10 inclusive (unit length). This calibration base is enriched by measurements of energy spectra of shims of different thicknesses of calcium (Zeff = 20, points 701), of sulfur (Zeff = 16, points 702), and of beryllium (Zeff = 4, points 703).
    [0032] The invention is not limited to a decomposition in a base of two calibration materials. Decomposition can be carried out in a base of more than two calibration materials, for example three, four or more. In addition, the calibration materials considered do not have to check any particular condition. The use of bases with more than two materials may prove useful when there is a continuity of the energy spectrum (called "k-edge") in the measured energy range. The method according to the invention can comprise, after the estimation of the characteristic thicknesses, an additional processing step, for example comprising the estimation of a concentration in the sample, or an estimate of the effective atomic number of the sample. sample. In particular, it is possible to determine a function f such that: = f (Lech Lech Lezef) pii022, pme) (14) 5 are M calibration materials, pi the density of the calibration material j, the length estimation characteristic associated with the calibration material j.
    [0033] For example: Lech f (x) = a + bx + cx2 + dx3, where x = (15) PiLeich + 102 -Le2ch With a, b, c, d are real and for example MATI of polyethylene, MAT2 of PVC. If the base contains more than two materials, the following function can be used: f (pi, Leich) = Ereicheff. ipiLEr piLeich (16) p being estimated from measurements made on known Zeff materials.
    [0034] The invention is particularly applicable in the medical field, particularly for the analysis of biological samples by spectral tomography. The invention is not limited to the examples which have just been developed, and it will be possible to imagine numerous variants without departing from the scope of the present invention, for example other calibration materials, other types of interpolations. etc. For example, other definitions of the likelihood function may be considered, based on other statistical modeling of the transmission rate across the sample in the spectrometer.
    权利要求:
    Claims (14)
    [0001]
    REVENDICATIONS1. A method of characterizing a sample (402), by estimating a plurality of so-called characteristic thicknesses, each associated with a so-called calibration material, characterized in that it comprises the following steps: acquisition (31; 61) d an energy spectrum, said transmitted spectrum of the sample (Sech), said spectrum being defined by a number of photons transmitted through the sample in each channel of a plurality of energy channels located in a band X and / or gamma spectral; for each of a plurality of so-called calibration spectrums (sbase L2)), calculating (32; 62) the value of a likelihood function from said calibration spectrum (e (1se (Li; L2) ) and the transmitted spectrum of the sample (Sech), each calibration spectrum (sbase (Li; L2)) corresponding to the transmitted spectrum of a stack of shims, each shim being formed of a known thickness of calibration material; determining (33; 66,69) estimates of the characteristic thicknesses cgeb; eh.} associated with the sample from said values of a likelihood function and according to the maximum likelihood criterion.
    [0002]
    2. Method according to claim 1, characterized in that the determination (33; 66, 69) of the estimates of the characteristic thicknesses (LIch; Lr) associated with the sample comprises a search for a maximum of the values of the function of likelihood, the thicknesses associated with this maximum forming the estimates (7 ", 4" ") of the characteristic thicknesses (Le, Le2ch).
    [0003]
    3. Method according to claim 1 or 2, characterized in that it comprises at least one step of interpolation of the values of the likelihood or interpolation function of the calibration spectra. 3037401 28
    [0004]
    4. Method according to claim 3, characterized in that at least one interpolation step (66; 68) implements a nonlinear interpolation function. 5
    [0005]
    5. Method according to claim 3 or 4, characterized in that it comprises the following steps: interpolation (65) of the values of the likelihood function by a likelihood interpolation function dependent on at least one variable, each variable corresponding to the thickness of a calibration material; and - searching (66) for a maximum of the values of said likelihood interpolation function, the thicknesses associated with this maximum forming the estimates of the characteristic thicknesses. 15
    [0006]
    6. Method according to any one of claims 3 to 5, characterized in that it comprises the following steps: interpolation (68) of the calibration spectra by an interpolation function spectra dependent on at least one variable, each variable corresponding to the thickness of a calibration material; and 20 - searching for a maximum of the values of said interpolation function of the spectra, the thicknesses associated with this maximum forming the estimates of the characteristic thicknesses.
    [0007]
    7. Method according to any one of claims 3 to 6, characterized in that the following steps are implemented: interpolation (65) of the values of the likelihood function by a likelihood interpolation function, said values being associated with combinations of known thicknesses of calibration materials such that for each calibration material the associated thicknesses are in a respective first range; searching for a maximum of the values of said likelihood interpolation function, the thicknesses associated with this maximum forming approximate values (V1, V2) of the characteristic thickness estimates; interpolating (68) the calibration spectra by an interpolation function of the spectra depending on at least one variable, each variable corresponding to the thickness of a calibration material and taking values in a respective second interval ([ L'111 L'12]; [L'21, L'22]), narrower than the first interval associated with the same calibration material and centered on the approximate value (Vi, V2) associated with the same calibration material; for each of the values of said spectral interpolation function, calculating (62) the value of the likelihood function and searching (69) for a maximum of said values of the likelihood function, the thicknesses associated with this maximum forming consolidated estimates (Leich; Len of characteristic thicknesses.
    [0008]
    8. Method according to any one of claims 3 to 7, characterized in that the following steps are carried out: interpolation (78) of the calibration spectra by an interpolation function of the spectra, said calibration spectra being associated with combinations of known thicknesses of calibration materials such as for each calibration material, the associated thicknesses are in a respective first range; for each of the values of said spectral interpolation function, calculating (72) the value of the likelihood function and searching (73) for a maximum of said values of the likelihood function, the thicknesses associated with this maximum forming approximate values (42), 42)) estimates of characteristic thicknesses; interpolation (75) of the likelihood function values by a likelihood interpolation function dependent on at least one variable, each variable corresponding to the thickness of a calibration material and taking values in a second interval respective ([421); 422]; [422; 422]), narrower than the first interval associated with the same calibration material and centered on the approximate value (g2), -42)) associated with the same calibration material; and searching (73) for a maximum of the values of said likelihood interpolation function, the thicknesses associated with this maximum forming consolidated estimates (ziech; ae2ch) of the characteristic thicknesses.
    [0009]
    9. Method according to any one of claims 1 to 8, characterized in that the likelihood function is determined from a statistical modeling of the transmitted spectrum of the sample (Sech), according to a Poisson law.
    [0010]
    10. Method according to any one of claims 1 to 9, characterized in that the likelihood function calculated from said calibration spectrum and the transmitted spectrum of the sample, is defined by: in (V (Sech, sbase Lei))) j = 1 V is the likelihood function, in is the natural logarithm, spase (Li, ..., Lm) + Z sjech in (sfase (k, ..., Lm))) l = 1 25 sec, / = vr 1.Sich is the transmitted spectrum of the sample, presenting] = R energy channels, 3037401 31 sbase = e_isipase LM) is the calibration spectrum at j = R energy channels of the combination of M associated calibration materials each at a respective thickness 1,1, ..., Lm, C is a constant. 5
    [0011]
    11. Method according to any one of claims 1 to 10, characterized by the use of at least one calibration spectrum (Sbase (Li; L2)) corresponding to the transmitted spectrum of an annex standard, this annex standard consisting of of a determined thickness of a reference material, and being associated with a combination of known thicknesses of calibration materials such that at least one thickness takes on a negative value.
    [0012]
    12. A method according to any one of claims 1 to 11, characterized by a step of manufacturing a calibration base comprising said calibration spectra (Sb-e (Li; L2)), comprising: measurements of the transmitted spectra each of a plurality of shim stacks, each shim being formed of a known thickness of calibration material; a measurement of the transmitted spectrum of at least one additional standard of a determined thickness of a reference material. , and associated with a combination of known thicknesses of the calibration materials such that at least one thickness takes a negative value; ranking of the set of spectra measured in a single database connecting a spectrum to a combination of thicknesses of the calibration materials.
    [0013]
    13. A method according to any one of claims 1 to 12, characterized by a calculation of the average effective atomic number of the sample, according to the estimates of the characteristic thicknesses (Leich;
    [0014]
    A device (400) for characterizing a sample (402), comprising an electromagnetic source (401) emitting in an X and / or gamma spectral band, and a detector (403) for measuring a transmitted spectrum (Sech) of the sample, said spectrum (Sech) being defined by a number of photons transmitted through the sample (402) in each channel of a plurality of energy channels, characterized in that it comprises a processor (404) ), arranged to implement the method according to any one of claims 1 to 13, and a memory (405) receiving the calibration spectra (Sbase (Li; L2)) and connected to the processor (404). 10
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3106864B1|2018-07-25|Characterisation of a sample by representation in a basis of materials
    FR2692672A1|1993-12-24|Method and device for generating a discrete mass spectrum by data deconvolution
    FR2961904A1|2011-12-30|METHOD FOR IDENTIFYING MATERIALS FROM MULTI ENERGY X-RADIOGRAPHIES
    WO1994028398A1|1994-12-08|Method and device for determining the attenuation function of an object with respect to the transmission of a reference material thickness
    EP3084406B1|2018-01-10|Method for measuring the effective atomic number of a material
    EP3132358A1|2017-02-22|Digital detection method
    EP2649436B1|2017-08-09|Method of extracting a prime scattering spectrum
    EP2726815B1|2019-09-04|Method and device for identifying a material by the spectral analysis of electromagnetic radiation passing through said material
    EP3645981B1|2021-08-04|Method for measuring a speed of a fluid
    WO2011120785A1|2011-10-06|Method for spectrometric analysis and related device
    EP1081505A1|2001-03-07|Digital correlator and correlation method for use in telemetry systems
    FR3038722B1|2019-06-14|METHOD FOR IDENTIFYING A MATERIAL
    Camhy-Val et al.1970|Absolute Measurement of Mean Lifetimes of Excited State~ Using the Method of Correlated Photons in a Cascade
    EP1831722A2|2007-09-12|Processing a radiation representative signal
    FR3034221A1|2016-09-30|METHOD AND DEVICE FOR DETECTING RADIOELEMENTS
    EP1348968B1|2008-04-16|Digital circuit for measuring signal power
    WO2006092520A1|2006-09-08|Method and system for physicochemical analysis using a laser pulsed ablation
    EP2769209B1|2019-01-02|Method and device for characterising a material by electromagnetic radiation scattering
    WO2020128284A1|2020-06-25|Method for characterising a spectrometer, computer program product and associated processor
    WO2016102861A1|2016-06-30|Method for estimating x-ray spectra of superimposed objects
    EP3635377A1|2020-04-15|Method for determining physical properties of a sample
    FR3089638A1|2020-06-12|OPTIMIZATION PROCESS, COMPUTER PROGRAM PRODUCT, ANALYSIS METHOD, AND RELATED CALCULATOR
    同族专利:
    公开号 | 公开日
    US20160363442A1|2016-12-15|
    EP3106864A1|2016-12-21|
    EP3106864B1|2018-07-25|
    US10969220B2|2021-04-06|
    US20210199429A1|2021-07-01|
    FR3037401B1|2017-06-23|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US20100202584A1|2009-02-06|2010-08-12|The Board Of Trustees Of The Leland Stanford Junior University|Optimal weights for measuring spectral x-ray data|WO2019136548A1|2018-01-09|2019-07-18|Voti Inc.|Methods for extending a range for assigning attributes to an object in an image|US6124141A|1998-01-07|2000-09-26|International Business Machines Corporation|Non-destructive method and device for measuring the depth of a buried interface|
    US7254501B1|2004-12-10|2007-08-07|Ahura Corporation|Spectrum searching method that uses non-chemical qualities of the measurement|
    US7352466B2|2005-06-17|2008-04-01|Canon Kabushiki Kaisha|Gas detection and photonic crystal devices design using predicted spectral responses|
    US9603249B2|2009-06-02|2017-03-21|Hsio Technologies, Llc|Direct metalization of electrical circuit structures|
    FR2953298B1|2009-11-30|2014-10-31|Commissariat Energie Atomique|METHOD OF CORRECTING THE STACK PHENOMENON APPLIED TO ACQUIRED X-RAY SPECTRA USING A SPECTROMETRIC SENSOR|
    FR2953603A1|2009-12-09|2011-06-10|Commissariat Energie Atomique|METHOD AND DEVICE FOR RECOGNIZING A MATERIAL USING ITS TRANSMISSION FUNCTION|
    FR2961904B1|2010-06-29|2012-08-17|Commissariat Energie Atomique|METHOD FOR IDENTIFYING MATERIALS FROM MULTI ENERGY X-RADIOGRAPHIES|
    FR2977321B1|2011-06-30|2015-01-23|Commissariat Energie Atomique|METHOD AND DEVICE FOR IDENTIFYING MATERIAL BY SPECTRAL ANALYSIS OF ELECTROMAGNETIC RADIATION THROUGH THIS MATERIAL|
    US8859969B2|2012-03-27|2014-10-14|Innovative Science Tools, Inc.|Optical analyzer for identification of materials using reflectance spectroscopy|
    US8929508B1|2012-04-17|2015-01-06|Robert E. Alvarez|Energy selective X-ray system with low noise|
    FR3015681B1|2013-12-20|2016-04-22|Commissariat Energie Atomique|METHOD OF MEASURING THE EFFECTIVE ATOMIC NUMBER OF A MATERIAL|EP3465180A4|2016-05-30|2020-03-04|Southern Innovation International Pty Ltd|Material characterisation system and method|
    FR3082945B1|2018-06-22|2020-06-05|Commissariat A L'energie Atomique Et Aux Energies Alternatives|METHOD FOR CHARACTERIZING AN OBJECT BY SPECTRAL IMAGING|
    法律状态:
    2016-07-08| PLFP| Fee payment|Year of fee payment: 2 |
    2016-12-16| PLSC| Search report ready|Effective date: 20161216 |
    2017-06-30| PLFP| Fee payment|Year of fee payment: 3 |
    2018-06-27| PLFP| Fee payment|Year of fee payment: 4 |
    2020-03-13| ST| Notification of lapse|Effective date: 20200206 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1555438A|FR3037401B1|2015-06-15|2015-06-15|CHARACTERIZATION OF A SAMPLE BY DECOMPOSITION BASED ON MATERIALS.|FR1555438A| FR3037401B1|2015-06-15|2015-06-15|CHARACTERIZATION OF A SAMPLE BY DECOMPOSITION BASED ON MATERIALS.|
    EP16174221.8A| EP3106864B1|2015-06-15|2016-06-13|Characterisation of a sample by representation in a basis of materials|
    US15/181,882| US10969220B2|2015-06-15|2016-06-14|Characterizing a sample by material basis decomposition|
    US17/185,092| US20210199429A1|2015-06-15|2021-02-25|Characterizing a sample by material basis decomposition|
    [返回顶部]