Method for spectroscopic determination and application of the method

Abstract

The invention discloses a method for the spectroscopic determination of at least one measured quantity in a medium, comprising the steps: - Definition of an application; - Providing a spectrometer; - Providing a raw spectrum, in particular by measuring the medium using a spectrometer; - Applying a calibration to the raw spectrum and obtaining a calibrated spectrum; - Standardizing the calibrated spectrum and obtaining a standardized spectrum, where standardization is carried out using spectrometer-type-specific information and a transformation function, wherein the transformation function includes the calibrated spectrum and additional information for the measurement as input variables, and includes a spectrometer-specific transformation function and calibration characteristics; - Selecting, if necessary adapting, or creating a spectrometer-independent and application-specific model from the standardized spectrum and application-specific data; and - Determining the measured quantity(s) from the raw spectrum using the model. The invention further relates to the application of the method in a Raman spectrometer.

Description

[0001] The invention relates to a method for the spectroscopic determination of at least one measurand in a medium. The invention further relates to an application of the method.

[0002] In spectroscopy, a "model" maps the raw spectrum, i.e., the pure measured values, to the measured quantity. For example, frequency-dependent cross sections are mapped to absolute concentrations of a specific substance in a medium. This applies to all spectroscopic methods, including Raman spectroscopy.

[0003] Such models are usually only valid for the instrument configuration used to collect the spectra for training the model. As soon as one or more components of the measuring instruments are replaced, the model becomes invalid and must be recreated, which often involves enormous experimental effort. Partially or completely repeating the experiments to record the training spectra requires a significant investment of time and potentially high costs, for example, in the biopharmaceutical sector, for providing the measurement medium in all the variations necessary for model creation, as well as personnel resources from appropriately trained staff and instrumentation. Creating a new model requires highly qualified experts with domain knowledge of the measurement medium, the spectroscopy method, and the model creation process itself.

[0004] The invention is based on the objective of simplifying model building.

[0005] The problem is solved by a method for the spectroscopic determination of at least one measurand in a medium, comprising the steps of: defining an application; providing a spectrometer; providing a raw spectrum, in particular by measuring the medium using the spectrometer; applying a calibration to the raw spectrum and obtaining a calibrated spectrum; standardizing the calibrated spectrum and obtaining a standardized spectrum, wherein the standardization is carried out using spectrometer-type-specific information and a transformation function, the transformation function comprising the calibrated spectrum and additional information for the measurement as inputs, and a spectrometer-specific transformation function and calibration characteristics;Selecting, if necessary adapting, or creating a spectrometer-independent and application-specific model from the standardized spectrum and application-specific data; and determining the measured quantity(ies) from the raw spectrum using the model.

[0006] The problem is solved by a method for the spectroscopic determination of at least one measurand in a medium, comprising the steps of: defining an application; providing a spectrometer; providing a raw spectrum, in particular by measuring the medium using the spectrometer; applying a calibration to the raw spectrum and obtaining a calibrated spectrum; selecting, if necessary adapting, or creating a spectrometer-type-specific and application-specific model from the calibrated spectrum, application-specific data, and by means of at least one database, wherein the database includes models for calibrated spectra and non-standardized spectra; and determining the measurand from the raw spectrum using the model.

[0007] This generally simplifies model building. For the user, the claimed method makes it possible to reduce costs and effort, thus making a spectroscopy technique that is normally very complex to set up, such as Raman spectroscopy, accessible even to users who do not have sufficient resources to create models themselves.

[0008] The claimed invention eliminates the need to retrain models for new instrument configurations, or at least significantly reduces the experimental effort required for model transfer. Additionally, the effort required for model adaptation (model structure, model weighting factors), verification, and validation is reduced. Furthermore, time is saved by eliminating or greatly simplifying the certification of the entire measurement process (for example, cGMP, the current Good Manufacturing Practice of the Food and Drug Administration (FDA)).

[0009] One embodiment provides that the raw spectrum is supplied from a source other than the measurement using the spectrometer. In one embodiment, this is obtained from the literature or by means of a different spectrometer (i.e., not the spectrometer used to measure the quantity in the medium). For example, the "other spectrometer" is a spectrometer of the same type / model. In another embodiment, the "other spectrometer" is a different type / model, particularly from a different manufacturer.

[0010] Logically, every spectrometer first acquires a raw spectrum. However, calibration from the raw data often takes place within the spectrometer itself, so the user does not have direct access to the raw spectrum. In this case, processing begins directly with the calibrated spectrum. The same often applies to spectra from the literature.

[0011] In one configuration, the raw spectrum output without an explicit calibration step fulfills the stability criteria that are also applied to the calibrated spectrum. In this case, the raw spectrum corresponds to the calibrated spectrum.

[0012] One implementation involves the qualitative, preferably quantitative, determination of the medium's composition, taking into account one or more substances to be determined, i.e., the measured quantity(ies). Specific examples include alcohol concentration in wine production, glucose or lactate concentration in cell culture media, or amino acids in the manufacture of pharmaceutical products. Further examples can be found downstream in pharmaceutical production, such as the processing of monoclonal antibodies (mAbs) or the separation of proteins and protein aggregates (high molecular weight species). The determination of methane, CO2, H2O, and H2S in natural gas is also possible.

[0013] One implementation involves calibrating the wavelength and / or the intensity. In Raman spectroscopy, for example, the Raman shift (usually in wavenumbers on the x-axis of the spectrum) and the intensity (usually in arbitrary units (au) on the y-axis) are plotted.

[0014] One implementation involves calibrating the measurement setup, including the spectrometer and any additional measuring equipment, particularly a probe. A calibration specific to this setup is applied to the raw spectrum, resulting in a calibrated spectrum. The calibration can consider either only the spectrometer or also additional measuring equipment (e.g., the probe, and potentially the process connection).

[0015] Standardizing the calibrated spectrum corrects any remaining differences (between spectrometers) after calibration. These can include, for example, systematic calibration errors, different resolutions (samples per wavenumber), different measurement ranges (e.g., in the Raman shift), different file formats of spectra from different instrument manufacturers, etc.

[0016] Ideally, a standardized spectrum for an application looks exactly the same, regardless of the measuring device used. Any remaining variations then stem solely from the application (excluding residual standardization errors).

[0017] It is also possible to account for statistical, non-systematic errors if the distribution is known. For this purpose, a set of standardized spectra is created from a single calibrated spectrum. This set of standardized spectra contains the expected variance due to the statistical dispersion around the assumed "true" spectrum. This set of standardized spectra is used in model training to make the model robust against the expected dispersions, so that a correct measurement can be achieved despite the dispersion.

[0018] One design provides that spectrometer-type-specific information includes explicitly provided or derived information from the spectrum, such as a reference spectrum like isopropanol, datasheet information, previous measurements, model, configuration, settings, etc.

[0019] The “spectrometer-specific transformation function” is a function that, taking into account the “spectrometer-type-specific information”, converts spectra from this instrument into standardized spectra.

[0020] For unknown instruments, the spectrometer-type-specific information would need to be so comprehensive that the transformation function could be created from it. If a sufficiently generalized transformation function is known (for example, because the spectrometer and application have been used in this combination multiple times), it can be configured and applied with less information.

[0021] One implementation involves using a distribution function in addition to the transformation function. This distribution function encompasses the distribution of the remaining calibration errors, where the standardized spectrum includes a probability of variations around the calibrated spectrum. The distribution function describes the distribution of these remaining calibration errors. In this case, the standardized spectrum is not a spectrum itself, but rather a statistically described probability of variations around the "true" calibrated spectrum. This variation may then also be a component of the standardized spectrum or the associated set of spectra.

[0022] One implementation provides that the transformation function is taken from a transformation function database, whereby the transformation function database is formed from already existing transformation functions.

[0023] One embodiment provides that the application-specific data includes information about the application, in particular the expected composition of the medium and the measured quantity(ies), the temperature, the pressure, and / or a background matrix of substances not to be determined. In one embodiment, the application-specific data includes a small number of spectra from measurements with known concentrations, with which the model is automatically fitted (e.g., a zero measurement, spans of 0% and 100%, etc.).

[0024] One implementation provides that new or adapted models are stored in a database, especially for models from standardized spectra, non-standardized and / or calibrated spectra, and / or where the models are loaded from a database.

[0025] One implementation involves converting standardized models into calibrated models and vice versa.

[0026] One approach involves using more than one model. In most cases, however, only one model will be used. If multiple models are employed to compare results in a "joint" analysis, possibly taking a quality metric into account, the best / most reliable result can be selected, or it is possible to combine the different results.

[0027] One implementation involves calculating and displaying a reliability value for the model. This uses the same or a similar metric as the one used to determine how well the model fits the model. In one implementation, concentration can be used at the model level. If the concentration falls within an acceptable range, the reliability is sufficient for this application. The measurement error of the concentration determination can also be used.

[0028] The problem is further solved by applying a method as described above using a Raman spectrometer.

[0029] This will be explained in more detail using the following figures. Fig. Figure 1 shows the claimed method. Fig. Figure 2 shows embodiments of some process steps of the claimed process. Fig. Figure 3 symbolically represents standardization.

[0030] Fig. Figure 1 shows the claimed method. Starting with an application, a spectrometer, for example a Raman spectrometer, is selected. The spectrometer serves for the spectroscopic determination of at least one measurand in a medium.

[0031] A spectrograph is an optical instrument that decomposes light of different wavelengths into its spectrum and records the resulting spectrum using suitable detectors. Instruments for spectroscopy, the visual observation of spectra, are called spectroscopes. A spectrometer is a device for displaying a spectrum. Unlike a spectroscope, it offers the possibility of measuring the spectra. In this document, "spectrometer" refers to the entire device that ultimately outputs the measured quantity(ies) as concentration values ​​from a spectrum using the model(s).

[0032] The spectrometer thus generates the raw spectrum of the medium. This raw spectrum may not be available to the user but is used solely for further processing. Usually, a calibration is applied to it (see below), and only this calibration is displayed to the user.

[0033] There are basically three possibilities, which are marked with the reference symbols 1, 2 and 3.

[0034] In all variations, more than a single spectrum is generally used. Through experimental design and execution, a set of spectra is created that represents the best possible compromise between effort and coverage of all relevant variances from the expected measurement. For the sake of simplicity, the following refers to "the" spectrum, but unless otherwise stated, this always means several spectra.

[0035] The left path, marked with the reference symbol "1", creates a model from the spectrum for the individual spectrometer and application. If this has already been done, it is selected. Thus, a model is generated specifically for this one spectrometer and this one application. The user then obtains the measured values ​​of the measured quantity from this model.

[0036] In the middle path, designated with the reference symbol "2", the raw spectrum is first calibrated. In one implementation, this is done using wavelength and intensity. From this calibrated spectrum, a model is created that is suitable for a specific spectrometer type and application. If this exact configuration already exists, the model is selected. If only minor details differ, the model is adapted. A corresponding database is also consulted in this process (see below).

[0037] The adaptation of the model can be of a device-side nature, for example, the replacement of a component with an identical component (e.g., replacing a probe with an identical probe).

[0038] The adaptation of the model can be of a process-related nature, for example through a very similar application (such as a changed composition of the nutrient solution in bioprocesses with essentially the same measurement task, or the same measurement task with slightly modified process control and corresponding variance in the spectra that was not previously covered by the model's training data set).

[0039] Adapting the model can include environmental parameters, such as application in a different factory, experimenter, different ambient air pressure / temperature, etc.

[0040] The adaptation of the model can be determined by the measured variables; for example, there is already a model for two measured variables, but the user wants three, two of which are the same as in the original model.

[0041] This results in a spectrometer-type-specific and application-specific model from which the measured quantity is determined.

[0042] The third path, 3, is described below. This path enables the best possible and simplest handling of the spectrometer. The idea is that the user only specifies what they want to measure. This method allows for measurements with reduced effort, whether for purely technical reasons and / or for certificates / audits.

[0043] First, the raw spectrum is calibrated as described above. Then, the calibrated spectrum is standardized. This is described below.

[0044] In the dashed path from the spectrometer to the standardizer, the raw spectrum, without an explicit calibration step, fulfills the stability criteria that are also applied to the calibrated spectrum. In this case, the raw spectrum corresponds to the calibrated spectrum, which is then standardized.

[0045] Finally, the model is recreated, selected, or adapted from the standardized spectrum, depending on the initial situation. The result is a model that is application-specific but independent of the spectrometer type.

[0046] Standardizing the calibrated spectrum corrects any remaining differences (between spectrometers) after calibration. These can include, for example, systematic calibration errors, different resolutions (samples per wavenumber), different measurement ranges (e.g., in the Raman shift), different file formats of spectra from different instrument manufacturers, etc.

[0047] Ideally, a standardized spectrum for an application looks exactly the same, regardless of the type of spectrometer used. Any remaining variations then stem solely from the application itself, and with successful standardization, their influence is small compared to the acceptable measurement error.

[0048] Fig. Figure 2 shows various procedural steps. The starting point is... Fig. 2 of a calibrated spectrum. It can be seen that there is a database for both the calibrated spectrum and the standardized spectrum, i.e., for path 2 and for path 3 (see Fig. 1) Each of these is referred to as a "Library." Depending on the type of spectrum, there is either one database per application and spectrometer type or one database per application. The databases can be converted into one another if the spectrometer-type-specific transformation function is known. With a known transformation function, the conversion of calibrated to standardized spectra is directly possible. The reverse conversion, from standardized to merely calibrated spectra, is only possible (losslessly) if the transformation function is uniquely reversible, taking into account all metadata for the respective spectrum.

[0049] “Metainformation” includes, for example, the analyte(s), background matrix, application, exposure time, integration time, temperature, pressure, hardware, manufacturer specifications, etc.

[0050] It is also possible to use artificial spectra. These can be either completely synthetically generated spectra or modified real spectra, so-called augmented spectra. An example of this is the aforementioned set of standardized spectra (for example, to account for statistical, non-systematic errors). This set contains the expected variance due to the statistical dispersion around the assumed "true" spectrum.

[0051] Data, i.e., spectra, from the databases can be used for each model. The selection of spectra from the database is such that, with or without additional measurement spectra, they allow for a sufficiently good model selection, modification, or creation for the measurement task.

[0052] There are databases for calibrated spectra and standardized spectra.

[0053] Furthermore, there are databases for the models, at least one database for models with non-standardized spectra and one with standardized spectra. There are one or more models for path 2 and for path 3, see the Fig. 2. Transformation between models is possible. A new model can be loaded into the database, while an existing model can be retrieved from the database. Without such a database, new training data (spectra) would have to be generated for each application. If the model already exists, it can be easily selected or optionally adapted for improved performance. If the model does not yet exist, it is created. The new model may be created based on existing models that closely resemble the model to be created. Thus, either spectra are retrieved from a database to reduce the effort required to provide spectra for model selection, adaptation, and creation, or one or more models are retrieved from a database to reduce the effort required for model adaptation or creation.

[0054] In one configuration, new or adapted models are loaded into the database.

[0055] The fact is that models for non-standardized spectra are usually not transferable to other spectrometer types with acceptable accuracy.

[0056] Ideally, the model is selected automatically and transparently for the user. This makes it as easy as possible for users to obtain acceptable measurement results. In addition to the aforementioned automated model selection, databases containing models can be included with the device or made available in the cloud / centrally accessible, for example, by the device itself, to further simplify the process.

[0057] Fig. Figure 3 symbolically illustrates the standardization process. Standardization is carried out using a "spectrum standardizer".

[0058] With perfect calibration, standardization is only necessary to standardize the resolution of the x- and / or y-axis. In most cases, however, standardization is also required to compensate for any residual error remaining in calibrated spectra.

[0059] If the calibrated spectrum already meets the requirements for a standardized spectrum, the "standardization" step may simply consist of forwarding the data. Only metadata might then be added.

[0060] After calibration, imperfections and variations may exist, including calibration tolerance, resolution, range, and signal-to-noise ratio (SNR). The standardizer cannot add missing information. A standardized spectrum cannot be better than the calibrated spectrum (unless additional information about the instrument is available) and can only make spectra equally "bad," not better. Noise can only be reduced to a limited extent. The information content of the spectra, which is relevant for model evaluation and the calculation of measurement results, must not be negatively affected by noise reduction. Resolution, however, can be adjusted if the measurement was not undersampled. The measurement range can be reduced. Systematic and known calibration errors can be compensated for.

[0061] The standardizer is supplied in a configuration that aligns with the model. Ideally, the user won't notice that the model pre-standardizes the spectra, but the model automatically becomes more robust and generic.

[0062] There can be one or more standardizers per spectrometer.

[0063] A standardized spectrum can be transferred from one analyzer to a second, and vice versa, using "inverse standardization." This involves transforming the standardized spectrum, originally measured with the first spectrometer type, into a format corresponding to a calibrated spectrum of the second spectrometer type using the inverse standardization function of that type. This allows models originally created for calibrated spectra of the second spectrometer type to also be used for spectra of the first. The reverse process, from the second to the first, is also possible. In both cases, the transformation function must be reversible.

[0064] The standardizer receives the calibrated spectrum, spectrometer-type-specific information, and a transformation function as input. The output is the standardized spectrum.

[0065] The spectrometer type-specific information consists of explicitly provided or spectrum-derived information about the spectrometer type, e.g., a reference spectrum, datasheet information, in-house measurements, etc.

[0066] The transformation function can also be stored in a database ("library") containing existing transformation functions.

[0067] The transformation function is derived from the calibrated spectrum and calibration characteristics. From this, the transformation function specific to the spectrometer type is extracted.

Claims

[1] Method for the spectroscopic determination of at least one quantity in a medium, comprising the steps: - Definition of an application; - Providing a spectrometer; - Providing a raw spectrum, in particular by measuring the medium using a spectrometer; - Applying a calibration to the raw spectrum and obtaining a calibrated spectrum; - Standardizing the calibrated spectrum and obtaining a standardized spectrum, where standardization is carried out using spectrometer-type-specific information and a transformation function, wherein the transformation function includes the calibrated spectrum and additional information for the measurement as input variables, and includes a spectrometer-specific transformation function and calibration characteristics; - Selecting, if necessary adapting, or creating a spectrometer-independent and application-specific model from the standardized spectrum and application-specific data; and - Determining the measured quantity(s) from the raw spectrum using the model. [2] Method for the spectroscopic determination of at least one quantity in a medium, comprising the steps: - Definition of an application; - Providing a spectrometer; - Providing a raw spectrum, in particular by measuring the medium using a spectrometer; - Applying a calibration to the raw spectrum and obtaining a calibrated spectrum; - Selecting, if necessary adapting, or creating a spectrometer-type-specific and application-specific model from the calibrated spectrum, application-specific data, and using at least one database, wherein the database includes models for calibrated and non-standardized spectra; and - Determining the measured quantity(s) from the raw spectrum using the model. [3] Method according to claim 1 or 2, wherein the calibration includes the wavelength and / or the intensity. [4] Method according to one of the preceding claims, wherein the calibration takes into account the measurement setup with spectrometer and connected additional measuring equipment, in particular a measuring probe. [5] Method according to one of the preceding claims, wherein the application is the qualitative, preferably quantitative, determination of the composition of the medium taking into account one or more substances to be determined, i.e. the measured quantity(s). [6] Method according to one of the preceding claims, wherein in addition to the transformation function a distribution function is used which includes the distribution of the remaining residual errors of the calibration, wherein the standardized spectrum includes a probability of variations around the calibrated spectrum. [7] Method according to one of the preceding claims, wherein the transformation function is configurable, wherein features of the standardized spectrum are adjustable, in particular the wavenumber range and the resolution of the output spectrum. [8] Method according to any of the preceding claims, wherein spectrometer type-specific information is explicitly provided or derived from the spectrum such as - a reference spectrum, such as isopropyl alcohol, - Datasheet information, in particular the resolution and measuring range in wavenumbers, - previous measurements, - the configuration, - Settings, - known and / or systematic calibration errors that can be subsequently compensated in the calibrated spectrum, in particular due to nonlinearities of the wavelength axis or intensity calculation, - Temperature dependencies, - Artifacts in the spectrum, especially from optical elements in the device or probe or the process connection, the electronics or the signal processing in the device - etc. [9] Method according to any one of the preceding claims, where the transformation function is taken from a transformation function database, where the transformation function database is formed from already existing transformation functions, and / or where a modified transformation function is stored in a transformation function database. [10] Method according to any of the preceding claims, wherein the application-specific data includes information relating to the application, in particular the expected composition of the medium and the measured quantity(s), the temperature and the pressure. [11] Method according to any of the preceding claims, wherein new or adapted models are stored in a database, in particular for models from standardized spectra, non-standardized and / or calibrated spectra, and / or wherein the models are loaded from a database. [12] Method according to any of the preceding claims, wherein a reliability value is calculated and displayed for the model. [13] Application of a method according to one of the preceding claims in a Raman spectrometer.

Images