JP2014137350A - Method for removing noise of chromatogram, and chromatography apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To achieve a chromatogram noise-removing method capable of efficiently removing noise even in a minute region.SOLUTION: A periodic noise extraction part 2 extracts a periodic noise signal waveform from a chromatogram by using a noise extraction threshold, and a Fourier transformation part 3 performs Fourier transformation of the extracted periodic noise signal waveform and calculates an amplitude spectrum composed of amplitude and a frequency. A periodic noise reconstruction part 4 removes a frequency component less than a frequency threshold from the amplitude spectrum, calculates a revised amplitude spectrum, and reconstructs the periodic noise signal waveform on the basis of the calculated revised amplitude spectrum. A noise removal part 5 removes the reconstructed periodic noise signal waveform from the chromatogram.

Description

  The present invention relates to a noise removal technique for chromatograms obtained by liquid chromatography or the like.

  The waveform data detected by the liquid chromatograph is ideally derived only from the component to be analyzed. However, in general, a noise waveform is superimposed on a waveform derived from a component, which greatly affects detection of a target component. As an example, Non-Patent Document 1 describes that as “13 Data Quality Guarantee”, the magnitude of noise is obtained from noise waveform data and a detection lower limit is defined.

  When the post-column method is adopted as liquid chromatography, a relatively large periodic noise may be observed. If waveforms with a plurality of periods are superimposed and the sample component signals overlap there, detection will be hindered. Thus, a structural invention of the device has been made such as suppressing the pulsating flow of the pump, which is a cause of periodic noise generation (for example, Patent Document 1).

  In addition, in computer data processing, Non-Patent Document 2 describes a weighted moving average called Low-Pass, High-Pass, and Gaussian as a technique for smoothing fine noise included in a mass spectrum waveform. Yes. In order to remove noise having regularity, it is shown that a large peak seen in the result of Fourier transform of a spectrum waveform is removed and inverse Fourier transform is performed. That is, if a portion showing periodicity can be removed from the result of Fourier transform, periodic noise is removed from the spectrum waveform obtained by inverse Fourier transform, and the function as smoothing is introduced.

  The method by Fourier transform / inverse Fourier transform of Non-Patent Document 2 is also disclosed in Patent Document 2 as a means for removing noise from a liquid chromatograph. Here, as a data processing method for chromatography, a spectrum showing frequency and amplitude is obtained by Fourier transform of measurement data, and a frequency region near zero derived from baseline drift and a high frequency region derived from noise are excluded. The chromatogram which removed the influence of drift and noise by inverse Fourier transform is obtained.

  However, in the chromatogram obtained by inverse Fourier transform, there is a description that the low frequency region to which the peak contributes is also removed, so that the baseline is lowered (Patent Document 2). It is thought that the influence of the base line and the sample component is mixed in the result of the Fourier transform, and it is difficult to draw a line with a specific frequency value. Further, as a fundamental problem of the method by Fourier transform, it is shown that the base line of the reconstructed chromatogram is waved by the influence of the peak, so that the influence is excluded by fitting with a trigonometric function (Patent Document 2).

JP 2005-164510 A JP-A-62-32360

JIS K 0124 General Rules for High Performance Liquid Chromatography (Internet (URL: http: www.jisc.go.jp/app/JPS/JPS00020.html)) Prepared by the JPO General Affairs Department Planning and Research Section, Technical Trend Survey Group, Fundamental Technology Collection, Mass Spectrometry (Mass Spectrometry), May 12, 2006 (Internet (URL: http://www.jpo.go.jp) /shiryou/s_sonota/hyoujun_gijutsu/mass/mokuji.htm))

  However, as in the background art, the method of Fourier transforming waveform data detected by a liquid chromatograph, removing noise components from the result, and performing inverse Fourier transform, for example, in a trace region that is problematic at the lower limit of detection and the lower limit of quantification. There are challenges to apply. For example, in a post-column method liquid chromatography device, a pump that sends out a mobile phase and a pump that sends out a reaction reagent are used, and pulsation occurs between these two pumps, generating minute noise. In some cases, it has been difficult to remove this noise with the prior art.

  It is difficult to say that the prior art is sufficiently established as a technique for removing at least periodic noise found in measurement results of a liquid chromatograph.

  An object of the present invention is to provide a chromatogram noise removal method capable of satisfactorily removing noise even in a minute region near the detection lower limit in which periodic noise is remarkably superimposed on a waveform derived from a sample component. And realizing a chromatographic apparatus.

  In order to achieve the above object, the present invention is configured as follows.

  Extract the periodic noise signal waveform from the chromatogram, Fourier transform the extracted periodic noise signal waveform, calculate the amplitude spectrum consisting of the amplitude and frequency, calculate the frequency threshold for removing the signal component, A chromatogram denoising method and a chromatographic apparatus for removing a frequency component less than a threshold from an amplitude spectrum, calculating a revised amplitude spectrum, reconstructing a periodic noise signal waveform, and removing the waveform from the chromatogram.

  Realized chromatogram noise removal method and chromatographic apparatus that can remove noise well even in a very small amount of area near the lower limit of detection, where periodic noise is significantly superimposed on the waveform derived from the sample component. can do.

It is a figure which shows the internal structure regarding the noise removal process in the data processing part of one Example of this invention. It is a schematic block diagram of the liquid chromatography apparatus of a post column method. It is an internal block diagram of the periodic noise extraction part shown in FIG. It is an internal block diagram of a Fourier-transform part. It is a figure which shows an example of an original waveform. FIG. 6 is a partially enlarged view of the waveform shown in FIG. 5. It is a figure which shows a process result waveform. It is an enlarged view of the processing result of the data of FIG. It is a figure which shows an example of the result of a discrete Fourier transform. It is a graph which shows the example of the amplitude spectrum which shows the strength of a frequency component. It is a figure which shows an example of the amplitude spectrum before and behind smoothing. It is a figure which shows the example of a periodic noise reconstruction waveform. It is a figure which shows an example of a parameter setting screen.

  Embodiments of the present invention will be described below with reference to the drawings.

  A liquid chromatograph rides a liquid flow called a mobile phase in a tube called a separation column, and multiple sample components to be analyzed move at a speed corresponding to each of them, so that each component is temporally moved. This is an analysis method that separates The sample component separated here is detected using characteristics such as a difference in light absorption between the mobile phase and the sample component. If the sample component has no difference in light absorption from the mobile phase, it is difficult to detect.

  Therefore, there is a technique called post-column method in which physical properties such as light absorption are changed by chemical modification of the sample components separated by the reaction reagent to facilitate the detection. In this case, periodic noise may be measured in a chromatogram showing a change over time such as the amount of absorbed light due to the influence of a pump for feeding a mobile phase and a pump for feeding a reaction reagent. The present invention is for removing periodic noise from such measurement data.

  FIG. 2 is a schematic configuration diagram of a post-column method liquid chromatography apparatus to which the present invention is applied.

  In FIG. 2, the liquid chromatography apparatus includes a first pump 10 that sends out a mobile phase, a sample injection device 11 that places sample components on the flow of the mobile phase, and a separation column for separating individual sample components at a moving speed. 12, a second pump 13 for sending a reaction reagent to the sample component separated by the separation column 12, a reaction coil 14 for mixing the sample component and the reaction reagent, and chemically modifying the sample component, and a reaction coil 14 includes a detector 15 that detects the sample component modified by 14, and a data processing unit 16 that accumulates and analyzes the amount of absorption of light detected by the detector 15. The data processing unit 16 includes a display unit, and the display unit includes a screen for performing input of a threshold value, waveform display, data display, and the like.

  Here, the noise removal processing in the present invention is performed in the data processing unit 16. In the example of FIG. 2, the configuration of two pumps is shown, but the present invention can be applied to a configuration of three or more pumps. The present invention is applicable to any situation where periodic noise is measured, and is not limited to the post-column method. Although the example of the amount of light absorption was shown as the object to be observed, the present invention can also be applied to measuring fluorescence or measuring with a mass spectrometer.

  FIG. 1 is a diagram illustrating an internal configuration related to noise removal processing in the data processing unit 16. In FIG. 1, a solid line indicates a flow of processing, and a dotted line indicates a data flow. The change over time in the amount of light absorption measured with a liquid chromatograph is shown as the original waveform 1. The data processing unit 16 includes a periodic noise extraction unit 2, a Fourier transform unit 3, a periodic noise reconstruction unit 4, a noise removal unit 5, and a noise removal waveform storage unit 6. The noise removal waveform 6 is obtained by performing noise removal processing by the noise removal unit 5 from the periodic noise reconstruction waveform 9 obtained by the periodic noise reconstruction unit 4 and the original waveform 1. Each of these processes will be described in detail below.

(1) Periodic noise extraction unit 2 (first step)
FIG. 3 is an internal block diagram of the periodic noise extraction unit 2 shown in FIG. FIG. 5 is a diagram showing an example of the original waveform, and FIG. 6 is a partially enlarged view of the waveform shown in FIG. FIG. 7 is a diagram showing a processing result waveform.

  In FIG. 3, the periodic noise extraction unit 2 includes a periodic noise characteristic analysis unit 21, a threshold setting unit 22, a periodic noise rough extraction unit 23, a reference position calculation unit 24, and a periodic noise extraction processing unit 25. With.

  In FIG. 5, the horizontal axis is time (minutes), and the vertical axis is an absorbance chromatogram indicating the amount of light absorbed. In a region where no peak or the like exists, periodic noise is greatly observed and looks like a band. The vicinity of 16 minutes and 31 minutes 30 seconds of the original waveform shown in FIG. 5 is enlarged and shown in (A) and (B) of FIG. 6, respectively. In FIG. 6A, at least three components are observed, and it is considered that one component is also observed in (B), but periodic noise is a peak derived from the component in the peak of (B). Superimposed. In this state, it is difficult to accurately determine the peak area and height.

  Therefore, the periodic noise characteristic analyzer 21 shown in FIG. 3 identifies the appearance area of periodic noise by paying attention to the minimum point of the original waveform. For example, the standard deviation (SD value) of the distance from the adjacent local minimum point is obtained. The threshold value setting unit 22 outputs a threshold value multiplied by a pre-specified magnification such as 3.3 × 2 times the SD value. The periodic noise rough extraction unit 23 determines whether or not it is a local minimum point that becomes periodic noise based on the threshold (noise extraction threshold) output from the threshold setting unit 22, and determines that the periodic noise is determined as periodic noise. Output as a coarsely extracted waveform. Further, the processing of the periodic noise rough extraction unit 23 is applied to the entire chromatogram.

  An example will be described with reference to FIGS. The crosses in the graphs of FIGS. 6A and 6B are local minimum points. Here, the distance between adjacent X marks is obtained. If the distance is less than the threshold value, the waveform data between the X marks is valid. When the distance is greater than or equal to the threshold value, the waveform data is obtained by complementing the space between the x marks of the preceding and following effective regions with a straight line as an invalid region. Waveforms obtained as a result of such processing are shown as periodic noise rough extraction waveforms in FIGS. 7A and 7B correspond to the waveforms of FIGS. 6A and 6B, respectively.

  As shown in FIG. 7A, the three peaks in which no periodic noise is seen in FIG. 6A are eliminated and only the noise portion can be left. On the other hand, since (B) in FIG. 6 includes periodic noise in the entire peak region, the waveform in (B) in FIG. 7 has the same waveform as in (B) in FIG.

  Here, as the determination of the periodic noise, attention is paid to the distance between the minimum points, but a method of focusing only on the value on the horizontal axis or the value on the vertical axis is also conceivable.

  Next, processing of the reference position calculation unit 24 will be described. The periodic noise shown in FIG. 7 includes sample component information and the like, as is evident from the waveform shown in FIG. For the purpose of extracting only periodic noise as much as possible and eliminating signals derived from sample components, a reference position for each time is obtained. Here, a moving average was performed on the waveform obtained as a result of the periodic noise determination using data for 30 seconds before and after, and used as a reference position. FIGS. 7A and 7B show examples of reference positions.

  The periodic noise extraction processing unit 25 subtracts the waveform at the reference position from the periodic noise rough extraction waveform shown in FIG. Enlarged views of the results are shown in FIGS. 8A and 8B. Although the waveform obtained here has periodic noise as the main component, the influence of the sample component remains somewhat. This waveform is used for Fourier transform, which will be described later, and is a source for reconstructing periodic noise. It is desirable to extract a waveform that is as close as possible to the original periodic noise.

(2) Fourier transform unit 3 (second step)
FIG. 4 is an internal configuration diagram of the Fourier transform unit 3. As shown in FIG. 4, the Fourier transform unit 3 in FIG. 1 includes a discrete Fourier transform unit 31, an amplitude spectrum calculation unit 32, an amplitude spectrum smoothing unit 33, a frequency threshold determination unit 34, and an amplitude spectrum revision unit. 35 and a periodic noise superimposing unit 36. Reference numeral 38 denotes a storage unit that stores a real part, an imaginary part, an amplitude spectrum, a frequency, and a phase that are Fourier transform results.

  In FIG. 4, first, the discrete Fourier transform unit 31 performs a discrete Fourier transform on the periodic noise extraction waveform to obtain a real part and an imaginary part. The waveform shown in FIG. 9A is the real part, and the waveform shown in FIG. 9B is the imaginary part.

  Furthermore, the amplitude spectrum etc. calculation part 32 calculates | requires an amplitude spectrum, and calculates the frequency and phase corresponding to each element. FIG. 10 is a graph showing an example of an amplitude spectrum indicating the strength of the frequency component. In the case of FIGS. 9 and 10, the horizontal axis of these graphs is the frequency (count / minute). In particular, FIG. 10 has a characteristic peak around 1 count / minute.

  In this example, two pumps 10 and 13 are fed at 0.5 mL / min and 0.6 mL / min, respectively, resulting in 5, 10, 15, 20, 25, and 30 counts / min. , 6, 12, 18, 24, 30, 36 counts / minute are superimposed on two series of periodic noises. Here, the series of frequency components that are integral multiples is due to the fact that the periodic noise derived from the pump is not a sine wave or cosine wave. Further, it is considered that the frequency region around 1 count / minute that is not serialized is not periodic noise.

(3) Periodic noise reconstruction unit 4 (third step, fourth step)
The processing operation of the periodic noise reconstruction unit 4 in FIG. 1 will be described with reference to FIG.

  In order to reconstruct periodic noise, it is necessary to obtain the frequency and amplitude corresponding to the periodic noise from the amplitude spectrum shown in FIG. 10 and reconstruct it in the original waveform format in which the horizontal axis represents time and the vertical axis represents absorbance. There is.

  In general, the amplitude and phase of the cosine wave are obtained from the values of the individual real part and imaginary part obtained by the discrete Fourier transform and superimposed, and the information corresponding to the periodic noise is retained from the result of the discrete Fourier transform. There is a method of Fourier transform. Here, periodic noise was reconstructed in the former.

  First, the amplitude spectrum smoothing unit 33 performs amplitude spectrum smoothing as preparation for obtaining a threshold value of a frequency for eliminating the above-mentioned around 1 count / minute. The amplitude spectrum that is the Fourier transform result is smoothed by the number of points indicated in the preset window value, and a smoothed amplitude spectrum is created. An example of the amplitude spectrum before smoothing is shown in FIG. 11A, and an example of the smoothed amplitude spectrum is shown in FIG.

  Next, the frequency threshold value determination part 34 calculates | requires the frequency equivalent to the tail of the peak around 1 count / min observed initially in a smoothed amplitude spectrum, and makes it a threshold value (frequency threshold value). Since the noise component is considered to have a higher frequency than the signal component, this threshold value is obtained in order to remove the signal component. However, in setting the threshold value, it is possible to adopt another standard different from the standard shown here in order to remove the signal component. An example of the threshold value is indicated by an arrow in FIG.

  Further, the amplitude spectrum revision unit 35 excludes a frequency region smaller than the threshold value from the amplitude spectrum and sets it as a revised amplitude spectrum (third step).

  FIG. 11C shows an example in which the threshold value 2.5 is applied to the amplitude spectrum shown in FIG.

  After obtaining the revised amplitude spectrum, the periodic noise superimposing unit 36 superimposes the waveform of the cosine wave specified from the corresponding frequency and phase for all amplitude values other than 0, and generates the periodic noise reconstructed waveform 37. (Fourth step).

  An example of the periodic noise reconstructed waveform obtained here is shown in FIG. Here, the waveforms shown in FIGS. 12A and 12B are obtained by processing the waveforms shown in FIGS. 8A and 8B.

  In one embodiment of the present invention, it is relatively easy to determine whether there is periodic noise in the amplitude spectrum. This means that in reconstructing periodic noise, there is little possibility of mixing sample component information, and noise can be accurately removed.

  Note that there is a method in which the frequency threshold for determining whether or not the noise is periodic noise is specified as a fixed value in advance. In addition, a method of setting a peak giving the slowest frequency in a series of peaks observed in the amplitude spectrum is also conceivable. For example, with a peak of 5 counts / minute as a reference, a method of setting a threshold by subtracting a specific value such as −2.5 counts / minute, or using a rising position of a peak of 5 counts / minute as a threshold.

  In addition, when the amplitude value indicated by the amplitude spectrum is small, the amplitude of the cosine wave is small and the contribution to the reconstructed periodic noise is small. Therefore, a method of copying only the portion constituting the series of frequencies taking an integral multiple to the revised amplitude spectrum is effective in terms of processing speed. For example, there is a method in which the standard deviation of the numerical value of the amplitude is obtained from the amplitude spectrum, the value is multiplied by a certain coefficient as the amplitude threshold value, and the amplitude obtained larger than the threshold value is copied to the revised amplitude spectrum. In that case, the range which calculates | requires a standard deviation is limited to 0-5 count / min etc., and the application of changing a threshold value by moving the area, etc. can be considered.

  In the determination of a series of frequencies that are integer multiples such as 5, 10,...

(A) Method for obtaining from peak determination result Individual peaks are determined from the amplitude spectrum, and a series in which the frequency is an integral multiple is obtained.

(B) Method of Deriving from Pump Control Information Periodic noise is caused by an internal repetitive operation when the pump sends liquid. The frequency of repetition, such as 5 counts / minute, is obtained from information such as the flow rate of the liquid. This determines the frequency series.

(4) Noise removing unit 5 (fifth step)
The noise removing unit 5 subtracts the reconstructed periodic noise from the original waveform as noise removal. 12A and 12B show waveforms after noise removal by the noise removal unit 5. The waveform after noise removal is stored in the noise removal waveform storage unit.

  In this way, it was possible to obtain a waveform from which periodic noise was largely excluded. The conventional method of excluding noise by moving average or the like is likely to have an effect such as the broadening of the peak width and the reduction of the height. In addition, the conventional method using Fourier transform has an influence such as excluding the sample component.

  On the other hand, in the case of the noise removal method according to one embodiment of the present invention, the influence can be reduced as much as possible.

  According to the conventional technique, in order to exclude a noise waveform, a waveform derived from a sample component is derived using Fourier transform / inverse Fourier transform. In the first place, in order to derive a waveform having no periodicity, the drawback of the discrete Fourier transform was an obstacle.

  In one embodiment of the present invention, the Fourier transform and the inverse Fourier transform are specialized for noise reconstruction, thereby avoiding defects and improving accuracy. Prior to the Fourier transform, there is a feature in that waveform data having a main component of periodic noise, which is compatible with the Fourier transform, is obtained.

  Here, in the present invention, it is also possible to calculate the periodic noise reconstructed waveform by performing inverse Fourier transform on the revised amplitude spectrum in the fourth step.

  The present invention is not necessarily limited to the case of the post-column method, and can also be applied to a pre-column method chromatography apparatus. For example, the present invention can be applied to a chromatogram in which periodic noise generated in a gradient liquid feeding method in which a plurality of pumps change the composition of a plurality of types of eluents to feed.

  Note that the present invention is not necessarily limited to liquid chromatography, and can be applied to gas chromatographs as long as they have a chromatogram superimposed with periodic noise. Further, the detector is not particularly limited as long as it outputs waveform data according to the chromatogram including the mass spectrometer.

  FIG. 13 is a diagram illustrating an example of a parameter input screen assuming that periodic noise is excluded in the data processing unit 16 according to the present invention. The meaning of each input item on the input screen shown in FIG. 13 is as follows.

  “Select Noise Threshold” is a threshold of the distance between valleys of a noise waveform for determining whether or not the noise is periodic noise. Above this value, it is not considered as periodic noise. Note that SD in the screen is a standard deviation of the distance between the valleys of the waveform, and is obtained from, for example, an area of 0 to 3 minutes of the original data. The variation is regarded as a normal distribution, and in this example, 2.0 times 3.3SD is set as a threshold value, and the operator can select how many times 3.3SD is set.

  “Baseline Calc Window (min)” is a moving average range for obtaining a reference position of periodic noise. In this example, a range of 1.0 minute is specified. In the screen example shown in FIG. 13, the reference position of each time in the chromatogram is called “Baseline”, but it does not mean the baseline of the sample component generally called, but is the Baseline as a reference of periodic noise. . “Baseline Calc Window (min)” can be changed by the operator.

  “DFT Select” is, for example, a threshold value for extracting a frequency component having a characteristic large amplitude in an amplitude spectrum as shown in FIG. In the example shown in FIG. 13, the standard deviation of the amplitude is SD, which is 1.5 times 3.3SD. The standard deviation is obtained while the range is expanded in the direction of 0 count / min from the amplitude of 75 count / min, which is the center of the amplitude spectrum. In addition, a standard deviation is obtained by determining a specific frequency range such as 10 counts / minute around the frequency to be determined.

  By providing a threshold in this way, it is possible to reconstruct a more efficient periodic noise by extracting frequency components having characteristic amplitudes such as 5, 6, 10, 12,... Is possible. 3.3 The number of times of SD can be selected by the operator.

  “Frequency” is a frequency range for obtaining a frequency component of periodic noise. The range shown in FIG. 13 is 3.5 to 75.0 counts / minute. In particular, in the case of FIG. 10, since the peak seen at 0 to 3.5 counts / minute is considered to be other than the periodic noise, it is set so as to exclude this portion. The last 75.0 counts / minute is the maximum frequency.

  As described above, according to the present invention, it is possible to remove only periodic components from the waveform data of the liquid chromatograph and facilitate detection of trace components derived from the sample.

  In particular, in the present invention, the periodic noise extraction waveform is subjected to discrete Fourier transform, the amplitude spectrum is smoothed, the bottom of the first peak in the direction in which the frequency increases from 0 to the threshold, and the amplitude spectrum before smoothing is The frequency region below this threshold is removed to remove the signal component. Thereby, it is possible to obtain a periodic noise waveform from which signal components have been successfully removed, and to calculate a signal waveform from which only noise components have been removed from the original waveform.

  The present invention is a basic technique for removing periodic noise observed in a liquid chromatograph, and in particular, improvement of the detection lower limit and the quantitation lower limit in the post-column method can be expected.

  For example, the BTB post column method, which is a kind of post column method, can contribute to highly sensitive analysis of organic acids such as formic acid, malic acid, lactic acid, acetic acid, and citric acid. Organic acids are targeted for analysis not only in foods but also in many fields such as pharmaceuticals, chemical industry, environmental analysis, and biotechnology. The OPA (orthophthalaldehyde) post-column method analyzes amino acids in foods and biological samples, the NIN (ninhydrin) post-column method analyzes proteins, peptides, and amino acids in biological samples, and the phenylhydrazine phosphate post-column method uses sugar. Analysis and application range is wide-ranging.

  When the detection lower limit and the quantification lower limit are improved by one digit, a trace component that is one digit thinner can be analyzed. Generally, when analyzing a trace amount component, means, such as extracting and concentrating the target component, are taken. For example, when 100 liters of river water is concentrated to obtain the sensitivity of the target component, if the detection sensitivity increases by an order of magnitude, the amount can be reduced to 1/10.

  It is even more compelling when analyzing biological samples such as experimental animals and humans, and it is often difficult to secure the amount of patient-derived diseased tissue necessary for analysis. Furthermore, it will spread to animal welfare issues such as how many model organisms such as mice are sacrificed.

  Therefore, since the present invention can easily detect a trace component derived from a sample, the amount necessary for the analysis is very small, and the burden on the living body can be reduced.

  DESCRIPTION OF SYMBOLS 1 ... Original waveform memory | storage part, 2 ... Periodic noise extraction part, 3 ... Fourier transform part, 4 ... Periodic noise reconstruction part, 5 ... Noise removal part, 6 ... Noise removal waveform storage unit, 9 ... periodic noise reconstruction unit waveform storage unit, 10 ... first pump, 11 ... sample injection device, 12 ... separation column, 13 ... second pump , 14 ... Reaction coil, 15 ... Detector, 16 ... Data processing unit, 21 ... Periodic noise characteristic analysis unit, 22 ... Threshold setting unit, 23 ... Rough periodic noise Extraction unit, 24 ... reference position calculation unit, 25 ... periodic noise extraction processing unit, 26 ... periodic noise extraction waveform, 31 ... discrete Fourier transform unit, 32 ... amplitude spectrum calculation, etc. Part, 33 ... amplitude spectrum smoothing part, 34 ... frequency threshold value judgment part, 35 ... Width spectrum revision unit, 36 ... periodic noise superimposing unit, 37 ... periodic noise reconstructed waveform, 38 ... Fourier transform result storing unit

Claims (11)

A first step of extracting a periodic noise signal waveform from a chromatogram using a noise extraction threshold;
A second step of Fourier-transforming the extracted periodic noise signal waveform to calculate an amplitude spectrum composed of amplitude and frequency;
A third step of calculating a frequency threshold for removing a signal component present in the calculated amplitude spectrum, removing a frequency component less than the calculated frequency threshold from the amplitude spectrum, and calculating a revised amplitude spectrum;
A fourth step of reconstructing a periodic noise signal waveform based on the revised amplitude spectrum;
A fifth step of removing the reconstructed periodic noise signal waveform from the chromatogram;
A method for removing noise from a chromatogram, comprising:   2. The chromatogram noise removing method according to claim 1, wherein the chromatogram is a chromatogram obtained by a liquid chromatograph.   2. The method for removing noise of a chromatogram according to claim 1, wherein in the first step, among a plurality of minimum points of the chromatogram, a distance between adjacent minimum points is less than a predetermined threshold distance between two points. A method for removing noise from a chromatogram, characterized by extracting a waveform to form a periodic noise signal waveform.   2. The chromatogram noise removal method according to claim 1, wherein the fourth step regenerates the periodic noise signal waveform by superimposing a cosine wave specified from the frequency and phase based on the revised amplitude spectrum. A method for removing noise from a chromatogram, characterized by comprising:   2. The chromatogram noise removal method according to claim 1, wherein the fourth step reconstructs a periodic noise signal waveform by inverse Fourier transforming the revised amplitude spectrum. Removal method. A liquid feeding section for feeding a liquid;
A sample injection part for injecting a sample into a flow path for supplying liquid from the liquid supply part;
A separation column for separating a sample component contained in the injected sample;
A detector for detecting a sample component separated by the separation column;
With
The data processing unit
Using a noise extraction threshold, a periodic noise extraction unit that extracts a periodic noise signal waveform from a chromatogram, and a Fourier transform that calculates an amplitude spectrum composed of amplitude and frequency by Fourier-transforming the extracted periodic noise signal waveform And a frequency threshold for removing a signal component present in the calculated amplitude spectrum, removing a frequency component less than the calculated frequency threshold from the amplitude spectrum, calculating a revised amplitude spectrum, and calculating the revised A chromatographic apparatus comprising: a periodic noise reconstructing unit that reconstructs a periodic noise signal waveform based on an amplitude spectrum; and a noise removing unit that removes the reconstructed periodic noise signal waveform from the chromatogram. Graphy equipment.   The chromatography apparatus according to claim 6, wherein the chromatogram is a chromatogram obtained by a liquid chromatograph.   The chromatographic apparatus according to claim 6, wherein the periodic noise extraction unit has a waveform in which a distance between adjacent minimum points among a plurality of minimum points of a chromatogram is less than a predetermined threshold distance between two points. A chromatographic apparatus characterized by extracting a periodic noise signal waveform.   7. The chromatographic apparatus according to claim 6, wherein the periodic noise reconstruction unit reconstructs a periodic noise signal waveform by superimposing a cosine wave specified from a frequency and a phase based on the revised amplitude spectrum. A chromatographic apparatus characterized by:   The chromatography apparatus according to claim 6, wherein the periodic noise reconstruction unit reconstructs a periodic noise signal waveform by performing inverse Fourier transform on the revised amplitude spectrum. A first pump for feeding the mobile phase;
A sample injection unit for injecting a sample into the mobile phase sent from the first pump;
A separation column for separating sample components from the mobile phase into which the sample has been injected;
A second pump for delivering a reaction reagent to the sample components separated by the separation column;
A reaction coil for mixing the sample component and the reaction reagent delivered by the second pump to chemically modify the sample component;
A detector for detecting a sample component modified by the reaction coil;
A data processing unit for analyzing the chromatogram detected by the detector;
The data processing unit includes:
Using a noise extraction threshold, a periodic noise extraction unit that extracts a periodic noise signal waveform from a chromatogram, and a Fourier transform that calculates an amplitude spectrum composed of amplitude and frequency by Fourier-transforming the extracted periodic noise signal waveform And a frequency threshold for removing a signal component present in the calculated amplitude spectrum, removing a frequency component less than the calculated frequency threshold from the amplitude spectrum, calculating a revised amplitude spectrum, and calculating the revised A chromatographic apparatus comprising: a periodic noise reconstructing unit that reconstructs a periodic noise signal waveform based on an amplitude spectrum; and a noise removing unit that removes the reconstructed periodic noise signal waveform from the chromatogram. Graphy equipment.

Images