JPS6232360A - Data processing method for chromatography - Google Patents

Abstract

PURPOSE:To reorganize a chromatogram from which the influence of a base line is eliminated by processing the chromatogram by utilizing frequency spectra. CONSTITUTION:The mixture sample obtd. by a chromatographic analysis is successively inputted via a detector 20, an A/D converter A/D and a peripheral apparatus connector 7 to a central processor 5. The input is converted to the data of the prescribed form by a control commander 6 to form the chromatogram having noise. A window function is fetched from a window function storage device 10 and the chromatogram having the noise is converted to the chromatogram for Fourier transform. The chromatogram is then converted to the spectra associated to frequencies in a fast Fourier transformer 9 and the spectra are stored 13. The spectra in the device 13 are subjected to reverse Fourier transform in the frequency region assigned by the transformer 9 and the window function is fetched from the device 10, by which the reorganized chromatograph for only the respective component peaks is obtd. The chromatogram having the influence of the noise and base line drift is subjected to the Fourier transform in the above-mentioned manner and is recognized to the chromatogram from which the influences thereof are eliminated.

Description

Detailed Description of the Invention [Field of Application of the Invention] The present invention relates to a data processing method for chromatography, and in particular to a data processing method suitable for processing data regarding a chromatogram obtained by chromatographically separating a mixture sample using a data processing system. Regarding the method.

[Background of the invention]

Chromatograms obtained by gas chromatography or liquid chromatography are often accompanied by noise or baseline fluctuations over the entire chromatogram. Baseline fluctuations in particular are caused by eluent gradient execution, temperature changes, etc. do. Further, noise that appears in the chromatogram is often caused by electrical noise, pump pulsation, or the like.

Methods to reduce the influence of baseline fluctuations and noise have been attempted for some time.
The issue is to eliminate the blank test results from the actual data to compensate for baseline drift.

Also, JP-A-48-55797 is the starting point of the chromatographic peak.

2. The end point is detected and the influence of noise is removed.

In conventional general data processing methods, baseline drift is processed assuming that it increases monotonically, but it is easily affected by noise and it is difficult to obtain a chromatogram from which baseline fluctuations are removed.

[Purpose of the invention]

An object of the present invention is to provide a chromatography data processing method capable of reconstructing a chromatogram from which the influence of the baseline has been removed.

[Summary of the invention]

The present invention focuses on the fact that the baseline drift is slower than the temporal change in the chromatographic peak, and processes a chromatogram using a five-frequency spectrum.

In the present invention, the chromatogram of a sample obtained by chromatographic analysis is digitally converted, the chromatogram data is then Fourier-transformed to obtain a frequency-related spectrum, and a specific frequency region of the spectrum is inversely Fourier-transformed. and outputting a chromatogram reconstructed by the inverse Fourier transform.

In a preferred embodiment of the invention, the above-mentioned specific frequency range is characterized in that it includes the frequencies of the spectrum originating from the peak, but does not include frequencies near zero originating from the baseline drift. Furthermore, the above-mentioned specific frequency range is characterized in that it does not even include a high frequency range derived from noise.

[Embodiments of the invention]

FIG. 2 is a diagram showing a hardware configuration applied to an embodiment of the present invention. In FIG. 2, ] is a signal A/D converter;
is a control command storage, 3 is a signal data storage, 4 is a parameter storage, and 5 is a central processor.

6 is a control command unit, 7 is a peripheral device connector, 8 is an output device,
9 is a fast Fourier transformer, 10 is a window function store, 11 is a chromatogram store, 12 is a chromatogram store for Fourier transform, 13 is a spectrum store, and 14 is a reconstructed chromatogram store.

In the mixture sample separated and analyzed by the chromatograph, each component peak reaches the detector 20 one after another, producing an analog signal as a chromatogram.

The analog signal from the detector 20 is converted to a digital signal via the A/D converter 1 and then via the peripheral device connector 7.
The central processing 815 is entered. The data is then converted into data in a predetermined format set by the control command device 6 based on the commands from the control command storage device 2, and stored in the signal data storage device 3. When the storage unit 3 reaches a certain amount of data, the output device 8 outputs the data.

This operation is repeated to obtain a noisy chromatogram as shown in FIG. The Fourier transform process is performed after the digitized chromatogram data is stored in the chromatogram storage device 11.

In response to a command from the control command unit 6, the window function is taken out from the window function storage 10, the stored chromatogram is made into a Fourier transform chromatogram, and the chromatogram is stored in the Fourier transform chromatogram storage 12.

Then, the fast Fourier transformer 9 converts the Fourier transform chromatogram into a frequency-related spectrum,
The spectrum is stored in the spectrum storage 13 and outputted to the output device 8. In accordance with the command from the control command unit 6, the spectrum in the spectrum store [3] is subjected to inverse Fourier transform in the frequency domain commanded by the fast Fourier transformer 9, and extracted into a window function from the window function store 10 as shown in FIG. Obtain a reconstructed romatogram of only peaks A and B. 2. This reconstructed romatogram is output to an output device and stored in the reconstructed romatogram storage 14.

After that, the parameters are taken out from the parameter storage 4 as before, and the central processor 5 performs peak detection and retention time 1.
Similar processing is performed when changing commands such as quantifying area and height.

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

FIG. 5 is a flowchart of one embodiment of the present invention, and is a flowchart of data processing suitable for mainly removing high frequency components from a 1M chromatogram.

In the figure, in step 30, the liquid chromatograph starts separating and analyzing the mixture sample, and in step 31, the signal from the chromatogram detector 20 in FIG. It is taken into the container 3. The acquisition timing, that is, the signal sampling interval is 0°1.
The operator can select from up to 10 seconds. The captured chromatogram signal is output as an original chromatogram to an output device 8 such as a CRT or printer in step 32. The chromatogram at this time is as shown in FIG. In step 33, the chromatogram data is stored in a chromatogram storage (D-RAM)6.Steps 34-38 are preparatory steps for the Fourier transform. When performing fast Fourier transform, the number of samples N per hour is n
must be 2n natural numbers (4-j), so a Fourier transform chromatogram is created before Fourier transform.

In step 34, the menu for the number of samples N is displayed on the CRT 8 in FIG. The number of samples is less than or equal to the sampling number of the original chroma 1-gram. In accordance with the output of this menu, the operator sets the number of samples N to the control command 6 in step 35.
Specify from. Step 36 is a step of dividing the chromatogram into N points, in which the chromatogram is extracted from the chromatogram storage 11' and is converted into a chromatogram with N samples by -order interpolation. In step 37, a window function, such as a Hanning function, is output from the window function store 10.

The values at each point of the chromatogram are multiplied by a window function, and the values at both ends of the time axis of the chromatogram are set to zero, thereby obtaining a Fourier transform chromatogram as shown in FIG. In the Fourier transform chromatogram storage step 38, the chromatogram shown in FIG.

Fourier transform the chromatograph related to retention time,
Obtain the frequency related spectrum. In step 40, the output device 8 displays the logarithmically expressed spectrum as the spectrum is output. In the storage step 41, the spectrum is stored in the spectrum storage 13. Step 42 is a step of specifying the frequency domain, and includes low frequency components near zero frequency derived from the baseline drift in the spectrum.

In order to remove high frequency components originating from noise, the operator specifies an intermediate frequency region using the control command device 6.

In the inverse Fourier transform step 43, the fast Fourier transformer 9 performs inverse Fourier transform only on the spectrum of the frequency domain specified in the step 42. Step 44 subjects the inverse Fourier transformed chromatogram to a window function in the window function store 10 to obtain a reconstructed chromatogram. Reference numeral 45 indicates the output of the reconstructed romatograms, and the reconstructed romatograms are smoothly connected, and the output device 8 outputs a reconstructed romatogram consisting of peaks A and B as shown in FIG. 4G is a storage process for the reconstructed romatogram, and the reconstructed romatogram is stored in the reconstructed romatogram storage unit 4. 47 is a peak detection process, in which the reconstructed romatogram is stored using the parameters in the parameter storage 4. Performs gram peak detection. Process 4
8 is a parameter in the parameter storage 4, and step 47
Quantify the retention time 1 area and height of the detected peak.

In step 49, it is determined based on a command from the control command unit 6 whether or not the designated frequency region is to be changed during inverse Fourier transformation. If there is a change, return to step 42. In step 50, it is determined based on a command from the control command unit 6 whether or not the number of samples N of the Fourier transform chromatogram is to be changed. If there is a change, the process goes through step 51 and returns to step 34. In step 51, when the chromatogram is output, the chromatogram in the chromatogram storage device 11 is taken out and outputted by the output device 8.

Step 52 is an analysis termination step, which is performed according to a command from the control command storage 2.

In this way, peaks could be detected and quantified even in a chromatogram with noise in low-frequency and high-frequency components.

According to this example, chromatograms that are affected by noise, pulsating flow, baseline drift, etc., which have conventionally caused incorrect peak detection and increased errors in quantitative determination of retention time, area, height, etc. The method using Fourier transform has the advantage of being able to reconstruct a chromatogram without these effects.

FIG. 7 is a flowchart showing another embodiment of the invention. In this example, the process from analysis start 30 to analysis end 52 is automated, and is configured to effectively remove high frequency noise.

In FIG. 7, steps having the same functions as those in FIG. 5 are given the same reference numerals. In the example of FIG. 7, after the yK chromatogram storage step 33, step 60 is executed. In this step 60, operations corresponding to steps 34 and 35 in FIG. 5 are automatically executed according to a predetermined program. That is, in order to execute step 60, the chromatographic retention time and the number of samples N are set in advance.

After the spectrum storage step 41 in FIG. 7, high frequency noise is recognized in step 62. This can be done by finding the frequency at which the amplitude of the spectrum goes to zero and then increases again. Then, in step 64, the region above that frequency is removed, and then in step 43, an inverse Fourier transform is performed on the remaining region from which the high frequency has been removed.

Here, we will briefly explain spectra and Fourier transform.

When the original chromatogram having baseline drift and noise as shown in FIG. 3 is subjected to Fourier transformation, a spectrum as shown in FIG. 8 is obtained.

The spectrum in FIG. 8 has three characteristic parts.

A is near the frequency zero, and 0, which is the part derived from the baseline drift, is the zero HE, which is derived from the peak.
This is the part that shows a Gaussian-like shape centered on . C is the part originating from high frequency noise. Of these,
Since A and C are interference components, select a frequency region so that these parts are removed (with the amplitude set to O),
When inverse Fourier transform is performed, a chromatogram containing only the peaks is obtained.

From the spectrum of FIG. 8, frequencies below 3 x 10-'Hyt and 1. When frequencies above OHZ are excluded, a spectrum as shown in FIG. 9 is obtained, and when this is subjected to inverse Fourier transform, a reconstructed romatogram free of noise and baseline drift is obtained as shown in FIG. 10. In the chromatogram shown in FIG. 10, the low frequency region to which the peak contributes is also removed, so the baseline is lowered by a certain value. In this case, by returning the baseline to zero, a reconstructed romatogram as shown in FIG. 4 can be obtained.

Since fast Fourier transform is performed in a finite interval, ripples appear and the spectrum from the baseline also widens.

In other words, it leaks into the low frequency range. For this reason, since the spectrum derived from the peak is also deleted, the baseline of the reconstructed romatogram becomes wavy as shown in FIG. 11.

Therefore, because the baseline and peak spectra overlap, a special method is needed to remove only the baseline.

When fast Fourier transform is performed with the spectrum in the low frequency region set to O, the waving function is a low frequency trigonometric function. The portion of the spectrum derived from the deleted peak is fit back using a trigonometric function, and a straight line is extrapolated to the chromatogram so that there are no peaks in the fitted section. Fitting of trigonometric functions only needs to be performed in the extrapolation interval; the straight line in the extrapolation interval becomes 0 in the chromatogram if low frequency components are sufficiently removed, so it does not affect the fitting. The baseline of the reconstructed chromatogram in the case of fitting is as shown in FIG.

FIG. 1 is a flowchart showing a preferred embodiment for applying the present invention to primarily eliminate baseline drift. 1st
In the figure, steps having the same functions as in the example of FIG. 5 are given the same reference numerals. In this example, after step 35 of specifying the number of samples N, in step 70, the surface edge of the chromatogram is extrapolated with a straight line. The degree of extrapolation is approximately 1000 times the half width of the peak in the chromatogram. Steps 37-41 are the same as in the example of FIG. In step 72, as shown in FIG. 9, the frequency region of the spectrum below 3.times.10-' is removed.

Then, in step 43, an inverse Fourier transform is performed, and in step 74, trigonometric functions in the extrapolated time domain are fitted in order to remove baseline distortion.

Then, in step 76, the baseline is corrected using the fitted parameters. After dividing the window function in step 44, the extrapolated time domain is deleted in step 78, and a reconstructed romatogram as shown in FIG. 4 is output.

Although omitted in FIG. 1, steps 48 and 52
In between, steps 49° and 50 and Sl are provided as in FIG. This embodiment allows accurate removal of baseline drift.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to remove baseline drift and noise that are obstacles when measuring peaks in a chromatogram.

[Brief explanation of drawings]

Figure 1 is a flowchart for explaining one of the embodiments of the present invention, Figure 2 is a diagram showing the hardware configuration applied to each embodiment of the present invention, and Figure 3 is an example of a chromatogram before Fourier transformation. FIG. 4 is a diagram showing an example of a chromatogram after reconstructing the chromatogram in FIG. 3, FIG. 5 is a flowchart explaining an embodiment of the present invention, and FIG. 7 is a flowchart explaining another embodiment of the present invention, FIG. 8 is a diagram showing an example of a spectrum obtained by Fourier transforming a chromatogram, and FIG. 9 is a flowchart explaining another embodiment of the present invention.
Figure 10 is a diagram showing an example of a reconstructed romatogram obtained by inverse Fourier transform of the spectrum in Figure 9, Figure 11 is a diagram showing an example of removing unnecessary frequency regions from the spectrum in Figure 9. FIG. 12 is a diagram showing a reconstructed chromatogram obtained by fitting a trigonometric function to the chromatogram of FIG. 11. 2... Control command storer, 3... Signal data storer,
4...Parameter storage unit, 5...Central processor, 6.
...Control command unit, 9...Fast Fourier transformer, 10.
... Window function storage, 14... Reconstructed romatogram storage.

Claims (1)

[Claims] 1. Digitally converting a chromatogram of a sample obtained by chromatographic analysis, and then Fourier transforming the chromatogram data to obtain a frequency-related spectrum; A data processing method for chromatography, comprising performing inverse Fourier transform on a region and outputting a chromatogram reconstructed by the inverse Fourier transform. 2. In the method according to claim 1, the specific frequency region includes the frequency of the spectrum originating from the peak, but does not include frequencies near zero originating from the baseline drift. A chromatography data processing method characterized by: 3. A data processing method for chromatography according to claim 2, wherein the specific frequency region does not include a high frequency region derived from noise.