Patent Application
20200335316
The embodiments herein generally relate to a system for optimizing peak shapes for a spectrometer, and, more particularly, to a system and a method for automatically optimizing peak shapes for a spectrometer such as a mass spectrometer for estimating gas mixtures.
The standard mass spectrometer produces a signature appearing at multiple mass to charge ratios (m/z ratios) associated with its ions and their fragments. The mass spectrometer may ionize different gases at different relative rates. Ions of the different gases may be fragmented and may appear at various mass to charge ratios (i.e. m/zs). The fragmented ions at various mass to charge ratios are transmitted to a detector. The fragmentation of the ion may be constant for one gas.
Mass spectrometer data typically shows “peaks” corresponding to individual ions with different mass to charge (m/z) ratios. The fragmentation of the ions may be obtained from a standard reference database or by experiment. Each peak of the fragmented ions typically includes a non-zero width, and possibly asymmetric shape which depends on the mass to charge ratio. The peak of the fragmented ions is varied between different classes of mass spectrometer instruments as the peak of the fragmented ions is specified based on the mass spectrometer. A perfectly ideal mass spectrometer has peaks of zero width (impulses), while every actual mass spectrometer shows peaks of non-zero width, and shapes varying from neat Gaussian or Lorentzian curves to combinations of multiple peaks curves overlapping each other.
In conventional mass spectrometers, each mass spectrometer employs an estimation algorithm for adapting to the peak shapes produced by the mass spectrometers. These mass spectrometers need an algorithm tuning steps where the algorithms implemented in each mass spectrometer is tuned to the specific peak shapes that a mass spectrometer produces. One of the approaches for shaping the overlapping peaks involves de-convoluting the shape of the overlapping peaks using a de-convolution process.
However, the de-convolution process fails to extract information from the minor peaks that are hidden under larger adjacent peaks. Moreover, this approach is an instrument specific calibration with a limited set of scaling factors. Further the above said approach has limited estimation accuracy, variations from unit to unit and limited sensitivity at higher mass to charge ratios. Said approach has been also adapted to other spectroscopic type sensors such as a Raman spectrometer, an absorption spectrometer or a vibrational spectrometer.
Accordingly, there remains a need for a system and a method that automatically optimizes any peak shapes for a mass spectrometer and other spectroscopic type sensors for estimating gas and other mixtures by automatically optimizing parameters of the sensors.
One of aspect of this invention is a system for estimating compositions of a target mixture using a first type sensor. The first type sensor generates a scan output for the target mixture. The scan output including spectra of detected compositions as a function of a first variable such as mass-to-charge ratio, wave number and others. The system comprises a data base and a set of modules. The data base stores characterization data of known mixtures, a set of constraints that includes accuracy, sensitivity and resolution required for an application to that the system applies, and an analytical model of a standard mixture. The set of modules comprises a peak shape identification module, a synthetic data pre-generation module, a cost function defining module, an actual peak shape generation module, a calibration module and an estimation module. The peak shape identification module is configured to identify a best peak shape for estimation of the compositions of the known mixtures such as know gas mixtures by analyzing the characterization data across the known mixtures, with added noise as a background of the application, wherein the best peak shape is referred as a peak shape meets the set of constraints of the application best. The synthetic data pre-generation module is configured to pre-generate synthetic data with a desired peak shape that is corresponding to the best peak shape from the analytical model with the standard mixture as input. The desired peak shape may be a peak shape of a part of spectra that has the same range of the best peak shape. The cost function defining module is configured to define a cost function to determine a peak shape that is suitable for estimation of the compositions of the target mixture from the best peak shape. The actual peak shape generation module is configured to generate a plurality of actual peak shapes, in the first type of sensor, for several different instances using the standard mixture to provide that an actual peak shape among the plurality of actual peak shapes as a calibrating input to calibrate the first type of sensor. The calibration module is configured to calibrate the first type of sensor by automatically adjusting parameters of the first type of sensor to find selected parameters for optimizing the actual peak shape to match with the desired peak shape. The estimation module is configured to estimate the compositions of the target mixture using the cost function from a peak shape of a scan output of first type sensor generating with the selected parameters.
In this system, the estimation module can estimate the compositions of the target mixture using the cost function from a peak shape of a scan output calibrated by the standard mixture without using de-convoluting the shape of the peaks included in the scan output.
The set of modules may further include a parameters validation module that is configured to validate the selected parameters by generating a scan output of a known mixture to estimate accuracy and peak shape quality. The best peak shape identification module identifies the best peak shape with added noise using machine learning.
The first type of sensor may generate a scan output for a target gas mixture, the scan output comprising the spectra of detected ions as a function of the mass-to-charge ratio corresponding to the target gas mixture. The calibration module calibrates the first type of sensor by adjusting the parameter comprises at least one of a Radio Frequency voltage to Direct Current voltage ratio, an Emission Current, voltage gradients and a bias voltage.
The calibration modules may include: (a) an optimizing module that is configured to optimize the parameters for a mass to charge ratio of interest once the parameters to be adjusted are selected; and (b) a determining module that is configured to determine each of the selected parameters is in a predefined range by constraining (i) optimization of the actual peak shape and (ii) optimization of each of the selected parameters to respective predefined range. The first type of sensor may include a mass spectrometer including a quadrupole mass filter. The selected parameter may include the voltage gradients and individual bias voltage comprising (i) box bias, (ii) Filament bias, (iii) Lens bias, (iv) Exit lens bias and (v) quadrupole bias.
The system may further comprise a memory that stores the database and the set of modules, and a processor that executes the set of modules. The system may further comprise a first type of sensor.
Another aspect of this invention is a method implemented on a computer that includes estimating compositions of a target mixture using a first type sensor. The first type sensor generates a scan output for the target mixture and the scan output includes spectra of detected compositions as a function of a first variable. The estimating composition includes: (a) identifying a best peak shape for estimation of the compositions of known mixtures by analyzing characterization data across the known mixtures, with added noise as a background of an application, wherein the best peak shape is referred as for a given set of constraints that includes accuracy, sensitivity and resolution in the application, a peak shape meets the set of constraints best; (b) pre-generating synthetic data with a desired peak shape that is corresponding to the best peak shape from an analytical model with standard mixture as input; (c) defining a cost function to determine a peak shape that is suitable for estimation of the compositions of the target mixture from the best peak shape; (e) generating a plurality of actual peak shapes, in the first type of sensor, for several different instances using the standard mixture to provide that an actual peak shape among the plurality of actual peak shapes as a calibrating input to calibrate the first type of sensor; (f) calibrating the first type of sensor by automatically adjusting parameters of the first type of sensor to find selected parameters for optimizing the actual peak shape to match with the desired peak shape; and (g) generating a scan output of the target mixture of the first type sensor with the selected parameters to estimate the compositions of the target mixture using the cost function from a peak shape in the scan output.
The estimating composition may further include validating the selected parameters by generating a scan output of a known mixture to estimate accuracy and peak shape quality. The step of identifying the best peak shape may include identifying the best peak shape with added noise using machine learning.
The first type of sensor may generate a scan output for a target gas mixture. The scan output may include the spectra of detected ions as a function of the mass-to-charge ratio corresponding to the target gas mixture. The step of calibrating may include calibrating the first type of sensor by adjusting the parameter comprising at least one of a Radio Frequency voltage to Direct Current voltage ratio, an Emission Current, voltage gradients and a bias voltage. The step of calibrating may include: (a) optimizing the parameters for a mass to charge ratio of interest once the parameters to be adjusted are selected; and (b) determining each of the selected parameters is in a predefined range by constraining (i) optimization of the actual peak shape and (ii) optimization of each of the selected parameters to respective predefined range.
The first type of sensor may include a mass spectrometer including a quadrupole mass filter and the selected parameter may include the voltage gradients and individual bias voltage comprising (i) box bias, (ii) Filament bias, (iii) Lens bias, (iv) Exit lens bias and (v) quadrupole bias.
The embodiments herein will be better understood from the following detailed description with reference to the drawings, in which:
The embodiments herein and the various features and advantageous details thereof are explained more fully with reference to the non-limiting embodiments that are illustrated in the accompanying drawings and detailed in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practiced and to further enable those of skill in the art to practice the embodiments herein. Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.
As mentioned, there remains a need for a system and a method that automatically optimizing peak shapes (i.e. Gaussian or Lorentzian curves or combinations of multiple peaks curves overlapping) for estimating a composition of a target mixture. The embodiments herein achieve this by providing an estimation system that generates an actual peak shape using standard mixtures to provide that actual peak shape as a calibrating input to calibrate the first type of sensor. Referring now to the drawings, and more particularly to
The estimation system 106 may be electrically connected to the first type of sensor 104. In an embodiment, the first type of sensor 104 includes a mass spectrometer sensor and/or spectroscopic type sensors (e.g. a mass spectrometer, a Raman spectrometer, an absorption spectrometer or a vibrational spectrometer). In an embodiment, one example of the first type of sensor 104 is disclosed in the U.S. Pat. No. 9,666,422. The first type of sensor 104 generates a scan output for a set of gases in the target gas mixture. The scan output includes spectra of detected ions as a function of the mass-to-charge ratio (a first variable) corresponding to the target gas mixture.
The target mixture 102a and the standard mixture 102b may be liquid mixtures, mixed solutions, mixed solids and others. The first type of sensor 104 may be other type of sensor such as a Raman spectrometer that generates a scan output includes spectra of detected compositions as a function of the wave number that is the first variable.
The estimation system 106 identifies a best peak shape for estimation accuracy of known gas mixtures by analyzing characterization data across the known gas mixtures, with added noise, using machine learning techniques. The best peak shape is referred as, for a given set of accuracy, sensitivity (i.e. minimum incremental concentration detectable) and resolution (i.e. distinguishing between similar ions (similar compositions)) constraints in the application to which the system 106 applies, a peak shape that can meet the constraints best. In an embodiment, the best peak shape is determined from the characterization data. The identification of the best peak shape includes obtaining the best peak shape for the estimation accuracy from the scan output of the first type of sensor 104 for the known gas mixtures. The characterization data refers scan outputs of the first type of sensor 104 from the same known gas mixtures at various parameters settings of the first type of sensor 104. In an embodiment, the parameter to an output shape relationship is varied from sensor to sensor.
The estimation system 106 pre-generates synthetic data with a desired peak shape from an analytical model with standard gas mixture 102b as input. The estimation system 106 further defines a cost function to determine a peak shape that is suitable for estimation of the target gas mixture 102a from the best peak shape. The estimation system 106 then generates a plurality of actual peak shapes in the first type of sensor 104 for several different instances using standard gas mixtures 102b to provide that an actual peak shape among the plurality of actual peak shapes as a calibrating input to calibrate the first type of sensor 104. In an embodiment, for each instance, the actual peak shape is generated based on different parameters of the first type of sensor 104. The estimation system 106 further calibrates the first type of sensor 104 by automatically adjusting the parameters of the first type of sensor 104 for optimizing the actual peak shape to match with the desired peak shape. In an embodiment, the parameter of the first type of sensor 104 includes at least one of a Radio Frequency voltage to Direct Current voltage ratio, Emission Current, voltage gradients and bias voltage. The voltage gradients and individual bias voltage parameter may include (i) box bias, (ii) Filament bias, (iii) Lens bias, (iv) Exit lens bias and (v) quadrupole bias. In an embodiment, the parameters of the first type sensor 104 are adjusted to effectively estimate desired peak shape of a particular gas in the target gas mixture. The estimation system 106 further validates the selected parameters including parameters that are specific to the mass to charge ratio of interest by generating a scan output of a known gas mixture 102c to estimate accuracy and peak shape quality. The estimation system 106 may be a computer, a mobile phone, a PDA (Personal Digital Assistant), a tablet, an electronic notebook or a Smartphone. In an embodiment, the first type of sensor 104 is embedded in the estimation system 106.
The peak shape identification module 204 identifies a best peak shape 204a for estimation of known gas mixtures by analyzing characterization data 202a across the known gas mixtures that are already analyzed by the first type of sensor 104. The peak shape identification module 204 identifies the best peak shape 204a with added noise, using machine learning techniques. The noise to be added is usually a background of spectral component of the application such as a spectral of an air, a carrier gas and others, e.g. noise of circuitries and amplifiers. In the peak shape identification module 204, the best peak shape 204a is referred as a peak shape meets the set of constraints 202b best.
The synthetic data pre-generation module 206 pre-generates synthetic data with a desired peak shape 206a from an analytical model 202c with the standard gas mixture 102b as input. The desired peak shape 206a corresponds to the part or the range of the best peak shape 204a in the spectral component of the pre-generated synthetic data of the standard gas mixture 102b. The cost function defining module 208 defines a cost function 208a to determine a peak shape that is suitable for estimation of the target gas mixture 102a from the best peak shape 204a. The actual peak shape generation module 210 generates a plurality of actual peak shapes, in the first type of sensor 104, for several different instances using standard gas mixtures 102b to provide that an actual peak shape 210a among the plurality of actual peak shapes as a calibrating input to calibrate the first type of sensor 104.
The calibration module 212 calibrates the first type of sensor 104 by automatically adjusting parameters of the first type of sensor 104 to find selected parameters 212a for optimizing the actual peak shape 210a to match with the desired peak shape 206a. In an embodiment, the parameters 212a to adjusted of the first type of sensor 104 includes at least one of a Radio Frequency voltage to Direct Current voltage ratio, Emission Current, voltage gradients and bias voltage. In another embodiment, the voltage gradients and individual bias voltage parameter includes (i) box bias, (ii) Filament bias, (iii) Lens bias, (iv) Exit lens bias and (v) quadrupole bias. The calibration module 212 includes a parameters optimization module 214 that optimizes the parameters for a mass to charge ratio of interest once the parameters 212a to be adjusted are selected. The calibration module 212 also includes a range determination module 216 that determines each of the selected parameters 212a is in a predefined range by constraining (i) optimization of the actual peak shape 210a and (ii) optimization of each of the selected parameters 212a to respective predefined range. The parameters optimization module 214 identifies the optimal parameters by the following equation.
Xn+1=Xn−K·Jcf(Xn),
Xn=nth set of parameters
K=constant
cf(X)=cost function
Jcf(X)=gradient vector of the cost function
The parameters optimization module 214 runs the gradient descent optimization over the selected parameters 212a to identify the optimal parameter. The parameters validation module 218 validates the selected parameters 212a including parameter that are specific to the mass to charge ratio of interest by generating a scan output of a known gas mixture 102c to estimate accuracy and peak shape quality. The estimation module 220 generates a scan output 220a of the target gas mixture 102a of the first type sensor 104 with the selected parameters 212a to estimate the compositions of the target gas mixture 102a using the cost function 208a from a peak shape in the scan output 220a.
At step 404, by the peak shape identification module 204, a best peak shape 204a for estimation of known gas mixtures is identified by analyzing characterization data 202a across the known gas mixtures, with added noise, using machine learning techniques. At step 406, by the synthetic data pre-generation module 206, synthetic data with a desired peak shape 206a is pre-generated from an analytical model 202c with the standard gas mixture 102b as input. At step 408, by the cost function defining module 208, a cost function 208a is defined to determine a peak shape whether that is suitable for estimation of the target gas mixture 102a from the best peak shape 204a. At step 410, by the actual peak shape generation module 210, a plurality of actual peak shapes are generated for several different instances in the first type of sensor 104 using standard gas mixtures 102b to provide that an actual peak shape 210a among the plurality of actual peak shapes as a calibrating input to calibrate the first type of sensor 104.
At step 412, by the calibration module 212, the first type of sensor 104 is calibrated by automatically adjusting parameters of the first type of sensor 104 to find selected parameters 212a for optimizing the actual peak shape 210a to match with the desired peak shape 206a. The parameter of the first type of sensor 104 to be adjusted includes at least one of a Radio Frequency voltage to Direct Current voltage ratio, Emission Current, voltage gradients and bias voltage. In an embodiment, the voltage gradients and individual bias voltage parameter includes (i) box bias, (ii) Filament bias, (iii) Lens bias, (iv) Exit lens bias and (v) quadrupole bias. In an embodiment, a stability of the system 106 is detected by determining whether the selected parameters 212a are within the allowable limits. The calibration 412 of the first type of sensor 104 may include steps of (a) optimizing the parameters for a mass to charge ratio of interest once the parameters to be adjusted are selected and (b) determining that each of the selected parameters is in a predefined range by constraining (i) optimization of the actual peak shape and (ii) optimization of each of the selected parameters to respective predefined range. At step 414, by the parameters validation module 218, the selected parameters 212a including parameters that are specific to the mass to charge ratio of interest are validated by generating a scan output of a known gas mixture 102c to estimate accuracy and peak shape quality.
A representative hardware environment for practicing the embodiments herein is depicted in
The estimation system 106 further includes a user interface adapter 19 that connects a keyboard 15, mouse 17, speaker 24, microphone 22, and/or other user interface devices such as a touch screen device (not shown) or a remote control to the bus 12 to gather user input. Additionally, a communication adapter 20 connects the bus 12 to a data processing network 25, and a display adapter 21 connects the bus 12 to a display device 23 which may be embodied as an output device such as a monitor, printer, or transmitter, for example.
The estimation system 106 is used to obtain better estimation accuracy from tall and thin peaks which are as close to Gaussian (normal) as possible. The estimation system 106 is used to minimize unit-to-unit (e.g. various mass spectrometers) variation. The estimation system 106 is used to tune the mass spectrometer 104 to various different applications (i.e. an ideal shape for each application is likely to be different and allow the mass spectrometer to be adapted).
One of the aspects of the above is a computer implemented system for optimizing a peak shape for estimating a composition of a target gas mixture, comprising: a first type of sensor 104 that generates a scan output for the target gas mixture, wherein the scan output comprises spectra of detected ions as a function of the mass-to-charge ratio corresponding to the target gas mixture; and an estimation system 106 that is connected to the first type of sensor 104 for estimating the composition of the target gas mixture. The estimation system comprises a memory that stores a database and a set of instructions, and a specialized processor that executes said set of instructions to (a) identify a best peak shape for estimation of known gas mixtures by analyzing characterization data across the known gas mixtures, with added noise, using machine learning, wherein said best peak shape is referred as, for a given set of accuracy, sensitivity and resolution constraints in an application, a peak shape meets the constraints best; (b) pre-generate synthetic data with a desired peak shape from an analytical model with standard gas mixture as input; (c) define a cost function to determine a peak shape that is suitable for estimation of the target gas mixture from the best peak shape; (d) generate a plurality of actual peak shapes, in the first type of sensor 104, for several different instances using standard gas mixtures to provide that an actual peak shape among the plurality of actual peak shapes as a calibrating input to calibrate the first type of sensor 104; (e) calibrate the first type of sensor 104 by automatically adjusting parameters of the first type of sensor 104 for optimizing the actual peak shape to match with the desired peak shape, wherein the parameter of the first type of sensor 104 comprises at least one of a Radio Frequency voltage to Direct Current voltage ratio, Emission Current, voltage gradients and bias voltage; and (f) validate the selected parameters comprising parameters that are specific to the mass to charge ratio of interest by generating a scan output of a known gas mixture to estimate accuracy and peak shape quality. Said calibrate comprises optimizing the parameters for a mass to charge ratio of interest once the parameters to be adjusted are selected; and determining that each of the selected parameters is in a predefined range by constraining (i) optimization of the actual peak shape and (ii) optimization of each of the selected parameters to respective predefined range.
The first type of sensor 104 may include a mass spectrometer. The voltage gradients and individual bias voltage parameter may comprise (i) box bias, (ii) Filament bias, (iii) Lens bias, (iv) Exit lens bias and (v) quadrupole bias.
In another aspect of the above, a computer implemented method for optimizing a peak shape for estimating a composition of a target gas mixture is provided. The method comprising: (a) generating 402, using a first type of sensor 104 a scan output for the target gas mixture, wherein the scan output comprises spectra of detected ions as a function of the mass-to-charge ratio corresponding to the target gas mixture; (b) identifying 404 a best peak shape for estimation of known gas mixtures by analyzing characterization data across the known gas mixtures, with added noise, using machine learning, wherein said best peak shape is referred as, for a given set of accuracy, sensitivity and resolution constraints in an application, a peak shape meets the constraints best; (c) pre-generating 406 synthetic data with a desired peak shape from an analytical model with standard gas mixture as input; (d) defining 408 a cost function to determine a peak shape that is suitable for estimation of the target gas mixture from the best peak shape; (e) generating 410 a plurality of actual peak shapes, in the first type of sensor 104, for several different instances using standard gas mixtures to provide that an actual peak shape among the plurality of actual peak shapes as a calibrating input to calibrate the first type of sensor 104; (f) calibrating 412 the first type of sensor 104 by automatically adjusting parameters of the first type of sensor 104 for optimizing the actual peak shape to match with the desired peak shape; and (g) validating 414 the selected parameters comprising parameters that are specific to the mass to charge ratio of interest by generating a scan output of a known gas mixture to estimate accuracy and peak shape quality. The parameter of the first type of sensor 104 comprises at least one of a Radio Frequency voltage to Direct Current voltage ratio, Emission Current, voltage gradients and bias voltage. Said calibrating comprises optimizing the parameters for a mass to charge ratio of interest once the parameters to be adjusted are selected; and determining that each of the selected parameters is in a predefined range by constraining (i) optimization of the actual peak shape and (ii) optimization of each of the selected parameters to respective predefined range.
In the above computer implemented method, the first type of sensor 104 may include a mass spectrometer. In the above computer implemented method, the voltage gradients and individual bias voltage parameter may comprise (i) box bias, (ii) Filament bias, (iii) Lens bias, (iv) Exit lens bias and (v) quadrupole bias. The above computer implemented method may further include the step of detecting a stability of the system by determining whether the selected parameters are within the allowable limits.
The foregoing description of the specific embodiments will so fully reveal the general nature of the embodiments herein that others can, by applying current knowledge, readily modify and/or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. Therefore, while the embodiments herein have been described in terms of preferred embodiments, those skilled in the art will recognize that the embodiments herein can be practiced with modification within the spirit and scope.
| Number | Date | Country | Kind |
|---|---|---|---|
| 201841000946 | Jan 2018 | IN | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/JP2019/000125 | 1/8/2019 | WO | 00 |
