JP5333329B2 - Chromatogram data processing method and apparatus - Google Patents

Description

  The present invention relates to a chromatogram data processing method and apparatus for processing chromatogram data obtained by chromatographic analysis such as gas chromatograph (GC) and liquid chromatograph (LC), and more specifically, a peak appearing in a chromatogram. The present invention relates to a data processing method and apparatus for correcting the holding time of the data.

  In the GC analysis, even if the analysis is performed with the same apparatus and under the same conditions, the retention time for the same component may be shifted due to various factors such as time variation of the carrier gas flow rate and column deterioration. is there. Therefore, in order to compare a plurality of chromatograms, it is necessary to correct the time axis so that the retention times for the same component are substantially the same prior to the comparison. If the hold time shift is completely linear, the time axis correction is easy, but generally the hold time shift is not small and has non-linearity. An algorithm based on dynamic programming (hereinafter referred to as “DP”) has been proposed as one of the time axis correction methods corresponding to such nonlinear holding time deviation (Non-Patent Document 1, Patent). Reference 1).

  The DP algorithm generally used in the past associates a reference signal (reference chromatogram signal) serving as a reference with a target signal (target chromatogram signal) whose time axis is distorted nonlinearly, and corresponds to the degree of distortion of the time. This is a method of searching for a correspondence relationship between the reference signal and the target signal that minimizes the calculated cost using the point intensity matching degree as a cost function. If this correspondence is obtained, the time axis of the target signal can be nonlinearly expanded / contracted by using this correspondence to correct the holding time shift.

Here, for the sake of explanation, it is assumed that the reference signal is A, and the sample point at each time of the reference signal A is represented by A (n) (where n is a positive integer). Similarly, the target signal is B, and the sample point at each time of the target signal B is represented by B (n). The search method for the optimal (best matching) correspondence in general DP is as follows (see FIG. 8).
[1] The sampling point of the target signal B corresponding to the sampling point A (1) of the reference signal A is, for example, “corresponding point” in consideration of a time-variable range (considering the maximum value of various variations). “None” may correspond to the target signals B (1) to B (3).
[2] Target to which the next sample point A (2) of the reference signal A can correspond depending on which of the above-mentioned range the sample point of the target signal B corresponding to the sample point A (1) of the reference signal A Each sample point group of the signal B is different. For example, if the sample point of the target signal B corresponding to the sample point A (1) of the reference signal A is B (1), the sample point group of the target signal B to which the next sample point A (2) can correspond is If there is no corresponding point or the target signal B (2) to B (4) and the sample point of the target signal B corresponding to the sample point A (1) of the reference signal A is B (2), then The sample point group of the target signal B to which the sample point A (2) can correspond is “no corresponding point” or the target signals B (3) to B (5).
Therefore, as shown in FIG. 8, as the number n of sample points A (n) increases, that is, as time elapses, the number of sample points that can correspond to the tree diagram increases.

Assuming that there are m candidates that can be dealt with in common sense when searching for the optimum correspondence as described above (m = 4 in the example of FIG. 8), all the chromatograms are obtained. If the data consists of data of n sample points, paths of approximately m to the power of n (that is, candidates for correspondences of the entire chromatogram) are derived. Therefore, the order of the number of candidates for the correspondence relationship between the reference signal A and the target signal B is O (m ⁿ ), and it is necessary to calculate a cost for each of them and select a candidate with the minimum cost.

  However, because there are restrictions on the amount of computation that can be processed due to the performance of the computer used for computation and the computation time, it is not realistic to search for all of the enormous number of candidates as described above and calculate the cost. Absent. Therefore, generally, a technique is adopted in which only the top x candidates are left in each search stage and the final number of candidates is reduced to x by deleting other data. Such a method is generally called beam limitation of the beam width x. The narrower the beam width x, the shorter the processing time is. However, if the beam width x is narrowed more than necessary, the matching in the first half of the search is good and the possibility of falling into a local solution with poor matching in the second half increases. Generally weakened by noise. On the other hand, in order to have resistance against such noise (robustness), it is necessary to allow a huge amount of time for arithmetic processing.

  That is, when a general DP algorithm as described in Non-Patent Document 1 and Patent Document 1 is applied to time axis correction of a chromatogram, an acquired signal with many contained components and a large number of peaks appearing in the chromatogram. Under bad conditions such as poor S / N and many false peaks or extremely large time fluctuations, the target signal cannot be properly matched with the reference signal in a realistic calculation processing time. There is a risk of incorrect time base correction. In particular, when the time variation accompanying the column exchange of GC analysis is large, or when analyzing a sample containing an enormous number of components such as gasoline and fragrance, the possibility that accurate time axis correction cannot be performed becomes considerably high.

International Publication No. 2004/090526 Pamphlet JP-A-5-181498

Pravdova and two others, "A comparison on two algorithms for warping of analytical signals", Analytical Kimika -Acta (Analytical Chimica Acta), 456, 2002, p. 77-92 Hiromitsu Miyazaki and two others, "Elastic matching algorithm for images based on dense DP", Image Recognition and Understanding Symposium (MIRU2004) Proceedings, July 2004

  The present invention has been made in order to solve the above-mentioned problems, and its main purpose is to perform appropriate processing even under adverse conditions such as a large number of contained components (peak number), a lot of noise, and a large time fluctuation. An object of the present invention is to provide a chromatogram data processing method and apparatus capable of correcting a time axis of a chromatogram with high accuracy within a time. The “chromatogram” mentioned here includes a total ion current chromatogram and an extracted ion chromatogram (mass chromatogram) obtained by a chromatograph mass spectrometer.

As a typical method for speeding up DP, the algorithm processing of DP is divided into two stages, a rough candidate search and a fine candidate search, so that both n and m in the number of candidates O (m ⁿ ) are reduced. Dense and dense DP is known. For example, Patent Document 2 discloses a technique using coarse / dense DP for speech recognition. Further, Non-Patent Document 2 proposes using coarse / dense DP for image pattern recognition involving deformation. In general, an audio signal or an image signal has a characteristic that a signal value at a certain position is highly correlated with a signal value at a position close in time or space. Therefore, it is relatively easy to apply the dense DP. On the other hand, in the case of a chromatogram signal, there is almost no correlation between a certain peak and other peaks that are close in time to it, and the above-described dense DP for voice or image cannot be used as it is. Therefore, the inventor of the present application introduced the method of coarse / dense DP to the time axis correction of the chromatogram, and applied the processing to reduce the calculation amount of DP as much as possible by using the characteristics of the chromatogram signal, to the present invention. I came up with an idea.

That is, the first invention made to solve the above problems is obtained by a chromatographic apparatus including a separation unit that separates various components contained in a sample in a time direction and a detector that detects the sample from which the components are separated. A chromatogram data processing method for correcting the time axis so that the time axis of the target chromatogram is aligned with the time axis of the reference chromatogram as a reference for the obtained chromatogram data,
a) using a peak detected on the reference chromatogram and the target chromatogram, and a linear correction step for removing a time variation having linearity from the target chromatogram;
b) Based on the intensity of the peak detected in the target chromatogram after the linear correction in the linear correction step and the reference chromatogram, the peak is selected for each of the reference chromatogram and the target chromatogram, A coarse search step for searching for candidates for the correspondence relationship between the reference chromatogram and the target chromatogram in the coarse stage by performing matching by a dynamic programming algorithm for the retention time of the peak,
c) For the candidate for the correspondence relationship in the coarse stage extracted in the coarse search step, after adding the peak excluded in the step, matching is performed by the dynamic programming algorithm for the peak retention time. A dense search step for searching for a correspondence relationship between the reference chromatogram and the target chromatogram in the dense stage;
d) a correction processing step for correcting the time axis of the target chromatogram based on the correspondence relationship between the reference chromatogram extracted in the dense search step and the target chromatogram;
It is characterized by having.

The second invention is an apparatus for carrying out the chromatogram data processing method according to the first invention, and a detection part for separating the components separated in the time direction and a detection for detecting the sample from which the components are separated. In the chromatogram data processing apparatus for correcting the time axis of the chromatogram data obtained by the chromatograph apparatus including the vessel so as to match the time axis of the target chromatogram with the time axis of the reference chromatogram serving as a reference,
a) linear correction means for removing time fluctuation having linearity from the target chromatogram using the peaks detected on the reference chromatogram and the target chromatogram;
b) Based on the intensity of the peak detected in the target chromatogram after the linear correction by the linear correction means and the reference chromatogram, the peak is selected for each of the reference chromatogram and the target chromatogram, A coarse search means for searching for a candidate for a correspondence relationship between a reference chromatogram and a target chromatogram in a rough stage by performing matching by a dynamic programming algorithm for holding the peak retention time;
c) For the candidates for the correspondence relationship in the coarse stage extracted by the coarse search means, after adding the peaks excluded by the means, matching is performed by a dynamic programming algorithm for the peak retention time. A dense search means for searching for a correspondence relationship between the reference chromatogram and the target chromatogram in the dense stage,
d) correction processing means for correcting the time axis of the target chromatogram based on the correspondence between the reference chromatogram extracted by the dense search means and the target chromatogram;
It is characterized by having.

  One of the biggest causes of the retention time shift that occurs in the chromatogram is column deterioration. The retention time shift due to this factor is a linear shift in time with some fluctuations. That is, the time axis of the target chromatogram is extended p times (or reduced to 1 / p) as a whole with respect to the time axis of the reference chromatogram, and further shifted by q overall. In the chromatogram data processing method and apparatus according to the present invention, the linear time lag described above is reduced in advance by the linear correction process in the linear correction step before executing the coarse / dense DP. As a result, in the subsequent dense / dense DP, it is only necessary to target non-linear holding time shifts, and even if the beam width is limited, that is, the search space is narrowed, matching is performed with sufficient robustness. Is possible.

  In the above-described general coarse / dense DP, first, measurement data is resampled to generate data with low time resolution, and this is used as input data to perform coarse-stage DP with a reduced number of data points (n). After that, m is reduced by applying the restriction that the optimum correspondence (matching) obtained under the low time resolution is not significantly violated, and then the original time resolution (before re-sampling) Perform DP processing. However, in the case of a chromatogram signal, the information itself of the original signal is lost by simple resampling. In the chromatogram signal, since the position (time) of the peak top of the peak is important as information, the detected peak position (time) is used as DP input data.

  However, when the number of contained components is large, the number of peaks also increases, and noise-derived peaks also appear in the chromatogram. Therefore, in the coarse search step, only the peaks that are likely to be particularly useful for correction are selected, for example, in order of intensity, and used as input data for the coarse DP, thereby reducing the number of data to be processed and simultaneously affecting the influence of noise. Reduce. Note that the peak height, peak area, or both can be used as the peak intensity used for peak extraction. Through the rough search step, candidates for correspondence between each extracted peak on the reference chromatogram and each peak extracted on the target chromatogram are searched. That is, for each extracted peak on the reference chromatogram, one or more peak candidates on the target chromatogram to be associated are listed.

  Next, in the dense search step, matching candidates are narrowed down by executing matching based on the candidates obtained in the coarse search step, including peaks that have been previously excluded due to low intensity. For example, for a peak on the reference chromatogram in which at least one corresponding peak candidate is listed in the coarse search step, if one of the candidates is listed as a candidate in the dense search step, that is the optimal solution. Discard other candidates as On the other hand, for the peak on the reference chromatogram for which there is no corresponding peak candidate in the coarse search step, the fine search is performed using the temporary position (time) obtained by linear interpolation of the candidate at the closest position in time. Cost calculation.

  The most probable correspondence between each peak on the reference chromatogram and each peak on the target chromatogram can be obtained by the two-stage DP of coarse and dense as described above. Thereafter, in the correction processing step, the time axis of the target chromatogram is corrected based on the correspondence relationship. Thereby, not only the position of the peak top but also the entire signal waveform of the target chromatogram is corrected based on the reference chromatogram.

  In the chromatogram data processing method according to the first aspect of the present invention, preferably, after execution of the coarse search step and before execution of the fine search step, a matching result that does not match the evaluation criterion is removed using the tendency of the entire matching as an evaluation criterion. It is preferable to have a filtering step. As a tendency of the entire matching, a statistical value of the entire variation amount may be used. For example, a standard deviation of the variation amount is obtained, and this standard deviation can be used as an evaluation criterion.

  According to this, it is possible to remove in advance those candidates that are not appropriate in view of the overall tendency from among the candidates for peak correspondence cited in the rough search step. Therefore, the number of correspondence candidates can be effectively reduced, the amount of calculation can be reduced, and the search accuracy in the dense DP can be increased.

  Moreover, as one aspect of the chromatogram data processing method according to the first aspect of the present invention, the linear correction step uses a correlation between the intensity of a plurality of peaks that are close in time and a time relationship to generate a reference chromatogram and a target chromatogram. For the linear correction by extracting the corresponding time range and checking the degree of matching of the peaks when performing linear correction assuming the combination of the time ranges and selecting the time range combination that shows the best match The coefficient can be obtained. Here, it is preferable to extract the time ranges corresponding to the first and last portions of the reference chromatogram and the target chromatogram, respectively.

  In addition, as a measure indicating the degree of coincidence of peaks in the linear correction step, for each peak on the reference chromatogram, one peak on the reference chromatogram and after linear correction is performed under the above assumption. On the target chromatogram, the sum of absolute values of time differences from the peak located closest to the peak intensity on the reference chromatogram and within the constant intensity ratio range can be used. Thereby, highly accurate linear correction can be performed by a comparatively simple calculation.

  According to the chromatogram data processing method and apparatus of the present invention, a coarse / dense DP improved so as to be suitable for the characteristics of chromatogram signals and the mode of expression of time fluctuations is used for correcting the time axis of the chromatogram, and Since correction for removing the linear time lag is performed before DP, the target chromatogram can be sufficiently matched with the reference chromatogram while efficiently reducing the number of correspondence candidates. As a result, for example, a highly accurate time axis within a practically sufficiently short time even under adverse conditions such as a large amount of components contained in the sample, a large number of peaks, a poor signal S / N, and a large time fluctuation. Correction can be performed. Thereby, for example, a plurality of chromatograms can be compared with high accuracy.

  In addition, since the data processing is sufficiently robust, various parameters can be used without changing parameters necessary for data processing such as weight of the cost function and beam width in the case of coarse / dense DP according to the type of sample and analysis conditions. Data obtained from various samples can be processed. Therefore, the burden on the operator during data processing is reduced, and the processing throughput can be improved.

1 is an overall configuration diagram of a GC apparatus to which a chromatogram data processing apparatus according to the present invention is applied. The flowchart which shows the process sequence of the time-axis correction method characteristic of this invention. The flowchart of the simple linear correction | amendment in FIG. Explanatory drawing of the concept of simple linear correction. Explanatory drawing of the process of simple linear correction. 3 is a flowchart of a rough stage DP in FIG. Explanatory drawing of the concept of the error candidate filtering process in FIG. Explanatory drawing of general DP. The figure which shows the simulation result for verifying the effect of the time-axis correction method characteristic of this invention.

  An embodiment of a gas chromatograph (GC) apparatus using a chromatogram data processing apparatus for performing a chromatogram data processing method according to the present invention will be described with reference to the accompanying drawings.

  FIG. 1 is a schematic configuration diagram of a GC apparatus according to the present embodiment. In FIG. 1, when a carrier gas is supplied at a constant flow rate from a sample introduction unit 1 including a sample vaporization chamber to a column 2, and the sample is injected into the sample introduction unit 1, the sample rides on the carrier gas flow to the column. 2 is introduced. Various components in the sample are separated in the time direction while passing through the column 2, and are sequentially eluted from the outlet of the column 2 and introduced into the detector 3. The detector 3 is, for example, a flame ionization detector (FID), a mass spectrometer, or the like, and outputs a detection signal corresponding to the amount of the introduced sample component every moment. This signal is converted into a digital value by the A / D converter 4 and input to the data processing unit 5. The data processing unit 5 includes a chromatogram data storage unit 51, a time axis correction calculation processing unit 52, a chromatogram creation / drawing unit 53, and the created chromagram is displayed on the display unit 6.

  The entity of the data processing unit 5 is a personal computer, and the functions of the above-described units can be realized by operating dedicated data processing software installed in advance in the computer.

  When GC analysis of the same sample is performed under the same analysis conditions using this GC apparatus and chromatogram data is acquired, ideally, the retention time for the same sample component is always the same. However, in practice, the change in the interaction between the sample component and the inner wall of the column 2 due to the deterioration of the column 2 over time, the temporal fluctuation of the flow rate of the carrier gas supplied from the sample introduction unit 1 to the column 2, The time axis fluctuates due to various factors such as the temporal fluctuation of the heating rate of the column oven (not shown), and the holding time for the same sample component is often not constant. This is particularly a problem when comparing multiple chromatograms. Therefore, in the GC apparatus of this embodiment, the time axis of the chromatogram acquired by GC analysis is corrected by the following time axis correction process.

  FIG. 2 is a flowchart showing a schematic procedure of time axis correction processing executed by the time axis correction calculation processing unit 52. The time axis correction processing here is based on the reference chromatogram stored in the chromatogram data storage unit 51, and the time axis of the target chromatogram obtained by GC analysis is matched with the time axis of the reference chromatogram. It is the process which correct | amends as follows. Basically, the reference chromatogram and the target chromatogram are obtained by analyzing the same type of sample.

  As a procedure for correction processing, first, peak detection is performed for each of the reference chromatogram and the target chromatogram, and the peak top position (time) and intensity (height, area, or both) are acquired as peak information. (Step S1). The time information of the peak detected here becomes input data in the coarse / dense DP described later.

  Next, in order to remove the linear holding time shift included in the target chromatogram, a simple linear correction process is executed (step S2). By this simple linear correction, linear components such as p-fold expansion and contraction in the time axis direction and time q shift are almost removed from the target chromatogram, and a retention time shift having nonlinearity remains. In practice, linear correction is performed on each peak on the target chromatogram, and the time information of those peaks may be corrected. Of course, it is possible to correct the linearity time lag in the coarse / dense DP, but usually the linearity time lag is much larger than the non-linear time lag, so both linearity and non-linear time lag are mixed. In the (superimposed) state, the search space for DP is considerably large. On the other hand, the search space can be narrowed by removing the linearity time shift in advance.

  After the simple linear correction, in each of the reference chromatogram and the target chromatogram, the peaks are selected based on the intensity information of each peak and reduced to a predetermined number. Then, the peak time information (the time information after linear correction for the target chromatogram) whose number is greatly reduced compared to the original number of peaks by this selection is used as input data, and matching based on the coarse DP algorithm is performed. Thus, a candidate for the correspondence relationship between the reference chromatogram and the target chromatogram is searched (step S3).

  After the candidate search by the coarse-stage DP is completed, before the next search by the fine-stage DP, the validity of the correction in terms of the overall tendency is selected among the candidates listed in the coarse-stage search by the filtering process. Then, matching that is estimated to be clearly wrong is removed (step S4). For example, a statistical value of the overall variation in time lag can be used as a criterion for determining whether or not there is an error. This filtering process can reduce search candidates.

  After that, using the matching result by the coarse stage DP, adding the peaks that were not selected by the above peak selection, that is, not considered in the coarse stage DP, and performing matching based on the dense DP algorithm As a result, the candidates for the correspondences listed in the rough stage DP are finally narrowed down to one (step S5). If an optimal correspondence of the target chromatogram with respect to the reference chromatogram, that is, the minimum cost is found by searching for candidates based on the dense DP, each sample point of the target chromatogram is obtained by warping based on the result. A signal that has been moved, that is, subjected to time axis correction is obtained (step S6). Thereby, chromatogram data in which the retention time deviation included in the target chromatogram is corrected is obtained.

  The above series of processes will be described in detail focusing on the processes characteristic of the present invention.

[Step S2] Simplified linear correction The main cause of the retention time shift of the chromatogram is the deterioration of the column 2, and the retention time shift due to this is almost linear with some fluctuations. That is, the general tendency of the time shift of the target chromatogram with respect to the reference chromatogram is linear. Therefore, by removing the linear holding time deviation before attempting the time axis correction by DP, the correction load by DP can be reduced and the correction accuracy can be improved. FIG. 3 is a flowchart showing the procedure of the simple linear correction process, FIG. 4 is an explanatory diagram of the concept of the simple linear correction, and FIG. 5 is an explanatory diagram of the process of the simple linear correction.

  As shown in FIG. 4, the time lag in the chromatogram is usually a superposition of a linear component and a nonlinear component, but the linear component is often dominant. When the linear component is not removed, the deviation amount that needs to be corrected by the coarse / dense DP is d2. However, if the linear component is removed, the deviation amount that needs to be corrected by the coarse / dense DP is only d1.

  The linear component of the time lag can be calculated from the start point and end point of the section where the peak exists on the chromatogram. For this purpose, it is necessary to detect the start point and end point with high resistance to noise and the like. Here, by evaluating the similarity between the reference chromatogram and the target chromatogram with a plurality of peaks appearing close in time as one group, the presence of peaks in the two chromatograms Detect the start and end points of the section. Then, the peak existing section of the target chromatogram is expanded and shifted in the time axis direction so that the start point / end point of the peak existing section of the target chromatogram matches the start point / end point of the peak existing section of the reference chromatogram.

  That is, first, in each of the reference chromatogram and the target chromatogram, leading and trailing candidates are extracted in a predetermined range near the start point and a predetermined range near the end point of the peak existing section (step S21). Specifically, as shown in FIG. 5, three temporally continuous peaks (triple peaks) are extracted, the correlation coefficient α of the intensity of the three peaks, the first peak and 2 A ratio β = T1 / T2 between the time interval T1 between the first peak and the time interval T2 between the second peak and the third peak is obtained. The correlation coefficient α and the time interval ratio β are parameters for determining the coincidence (similarity) between the triple peak on the reference chromatogram and the triple peak on the target chromatogram. And, the coincidence is evaluated by using the correlation coefficient α and the time interval ratio β for the triple peak on the reference chromatogram on the start point side and the triple peak on the target chromatogram to match. Select the one with the highest possibility as the tip candidate. Similarly, for the triple peak on the end point side on the reference chromatogram and the triple peak on the end point side on the target chromatogram, the one that is highly likely to match is selected as the last candidate.

  Next, one tip candidate (triple peak) and one tail candidate (triple peak) extracted in step S21 are selected one by one, and a linear correction is actually performed on the target chromatogram under the conditions. The degree of coincidence of peaks is determined in the target chromatogram and the reference chromatogram (step S22). The index for estimating the degree of coincidence between the peaks of both chromatograms is such that each peak appearing in the reference chromatogram has an intensity such that the ratio to the intensity of the peak falls within a certain range and is present at the closest position in time. The sum of distances (time) between the peaks on the target chromatogram to be used is used as an evaluation function. Peak matching evaluation is executed using the same evaluation function for all candidate combinations (step S3), and a combination of leading and trailing candidates that minimizes this evaluation function is selected (step S24). Actually, in order to simplify the calculation, an evaluation function obtained by adding the above-described evaluation function of the degree of coincidence and the evaluation function indicating the degree of beginning / end may be used.

  With the above processing, the start point and end point of the peak existing section are determined in the reference chromatogram and the target chromatogram, and thereby parameters for linear correction (specifically, the expansion / contraction rate p and the shift amount q) are calculated (step S25), the position of each peak on the target chromatogram is corrected according to the parameter (step S26). Thereby, the linear component of the time shift included in the target chromatogram is almost removed.

[Step S3] Search by Coarse Stage DP FIG. 6 is a flowchart showing a processing procedure of the coarse stage DP. In the coarse stage DP, in order to reduce the data to be processed, first, thinning is simply performed at a constant rate in order of peak intensity, and peaks whose intensity is below a certain relative intensity are considered to be likely to be noise and removed. Perform the process. As a result, the peaks that can be estimated to be significant for the correction are selected, and the number of peaks is suppressed to a predetermined number (step S31). Then, matching based on the DP algorithm is executed using the peak time of the reference chromatogram and the target chromatogram as input data. That is, with respect to the peak extracted on the reference chromatogram, the candidates for association with the peak extracted on the target chromatogram are listed in order from the earliest time, and when a plurality of candidates are listed, The search is repeated to list candidates for peak matching. Therefore, as shown in FIG. 8, candidates expand into a tree shape (step S32).

  At this time, the number of candidates listed at each peak point is limited by suppressing the spread of candidates to a time range that takes into consideration the maximum value of the assumed nonlinear time shift. As described above, since the linearity time lag has been removed in advance, even if this restriction is tightened (even if the beam width is narrowed), a situation where a valid candidate is excluded can be avoided.

Next, the cost is calculated for each of the searched correspondence candidates (step S33). The cost function at the coarse stage DP is as follows.
(1) Multiplying the logarithm of the intensity ratio of the corresponding peak by a constant. However, if there is no corresponding peak, it is assumed that a constant constant is added to the peak intensity for each skipped peak.
(2) Time variation: Absolute value of the time difference between the peak on the reference chromatogram and the peak on the target chromatogram (3) Difference in time variation: Most of the time variation occurs slowly, so the difference in time variation Should be close to zero. Therefore, as a value corresponding to the average of the time fluctuation amount, a value obtained by multiplying the time fluctuation amount by a low-pass filter and an absolute value of the difference in time fluctuation at the current corresponding peak are used.
(4) Difference in intensity ratio: There is a very strong correlation between the intensity of each peak in the reference chromatogram after matching and the target chromatogram, and the intensity ratio is ideally constant. Therefore, as a value corresponding to the logarithmic average of the intensity ratio, a value obtained by multiplying the logarithm of the intensity ratio by a low-pass filter and the absolute value of the logarithm of the intensity ratio at the current corresponding peak are used.
In practice, each of the items (1) to (4) is subjected to predetermined gamma correction and then multiplied by a constant to obtain a cost function.

  If the cost calculation has been completed for all candidates of the correspondence relationship between the reference chromatogram and the target chromatogram, the candidate with the lowest cost is selected and determined as the optimum matching of the rough stage DP (step S34).

[Step S4] Filtering Process Although DP is excellent for local matching, matching that deviates from the validity of correction in view of the tendency of the entire chromatogram may be performed. Therefore, as shown in FIG. 7, for the correspondence determined as the optimum matching, the standard deviation of the time deviation at each peak point is calculated, and if there is a deviation that deviates from the standard deviation, Remove matching.

[Step S5] Search by dense stage DP Next, after narrowing the search space using the matching result obtained in coarse stage DP, matching by DP algorithm including peaks excluded in coarse stage DP is also performed. Execute. At this time, since the time fluctuation in the fine stage DP takes a value close to the time fluctuation obtained in the coarse stage DP, the gamma correction and the constant applied to the cost function of the result of the coarse stage DP are changed. And the cost of the degree of divergence from the matching result at the coarse stage DP.
(1) For a peak on the reference chromatogram having a matching in the coarse stage DP, if the matching does not match in the dense stage DP, the cost is considered infinite, and the candidate is discarded.
(2) For peaks on the reference chromatogram that do not match in the coarse stage DP, the absolute value of the time lag with the time obtained by linear interpolation of the results matched in the coarse stage DP (peaks close in time) is the divergence degree cost. And

  As a result, it is assumed that the matching of the peaks matched in the coarse stage DP is correct, and the peaks that are not used in the coarse stage DP are also appropriately matched. As a result, the final chromatogram and the target chromatogram are obtained. It is possible to obtain the optimum correspondence with

  FIG. 9 is a diagram showing a processing result when time axis correction is executed by the chromatogram data processing method according to the present invention. As shown in this figure, it can be seen that the corrected target chromatogram is in good agreement with the reference chromatogram.

  The above-described embodiments are merely examples of the present invention, and it is a matter of course that changes, modifications, and additions are appropriately included in the scope of the claims of the present application within the scope of the present invention.

DESCRIPTION OF SYMBOLS 1 ... Sample introduction part 2 ... Column 3 ... Detector 4 ... A / D converter 5 ... Data processing part 51 ... Chromatogram data storage part 52 ... Time-axis correction arithmetic processing part 53 ... Chromatogram preparation and drawing part 6 ... Display Part

Claims (5)

Chromatogram data obtained by a chromatograph that includes a separation unit that separates various components contained in the sample in the time direction and a detector that detects the sample from which the components have been separated is displayed on the time axis of the reference reference chromatogram. A chromatogram data processing method for correcting the time axis so as to match the time axis of the target chromatogram,
a) using a peak detected on the reference chromatogram and the target chromatogram, and a linear correction step for removing a time variation having linearity from the target chromatogram;
b) Based on the intensity of the peak detected in the target chromatogram after the linear correction in the linear correction step and the reference chromatogram, the peak is selected for each of the reference chromatogram and the target chromatogram, A coarse search step for searching for candidates for the correspondence relationship between the reference chromatogram and the target chromatogram in the coarse stage by performing matching by a dynamic programming algorithm for the retention time of the peak,
c) For the candidate for the correspondence relationship in the coarse stage extracted in the coarse search step, after adding the peak excluded in the step, matching is performed by the dynamic programming algorithm for the peak retention time. A dense search step for searching for a correspondence relationship between the reference chromatogram and the target chromatogram in the dense stage;
d) a correction processing step for correcting the time axis of the target chromatogram based on the correspondence relationship between the reference chromatogram extracted in the dense search step and the target chromatogram;
A method for processing chromatogram data, comprising: The chromatogram data processing method according to claim 1,
A chromatogram data processing method comprising: a filtering step for removing matching results that do not conform to the evaluation criteria after the rough search step and before the dense search step as an evaluation criterion . The chromatogram data processing method according to claim 1 or 2,
In the linear correction step, a time range corresponding to the reference chromatogram and the target chromatogram is extracted using the correlation between the intensities of a plurality of peaks that are close in time and the time relationship, and a combination of the time ranges is assumed. Then, the degree of coincidence of peaks when linear correction is performed is determined, and a coefficient for linear correction is obtained by selecting a combination of time ranges showing the best match. The chromatogram data processing method according to claim 3,
As a measure indicating the degree of coincidence of peaks in the linear correction step, for each peak on the reference chromatogram, the target chromatogram after linear correction is performed under the assumption of one peak on the reference chromatogram. A chromatogram data processing method characterized by using a sum of absolute values of time differences with respect to a peak located within a certain intensity ratio range with respect to the peak intensity on the reference chromatogram on the gram. Chromatogram data obtained by a chromatograph that includes a separation unit that separates various components contained in the sample in the time direction and a detector that detects the sample from which the components have been separated is displayed on the time axis of the reference reference chromatogram. In the chromatogram data processing apparatus for correcting the time axis so as to match the time axis of the target chromatogram,
a) linear correction means for removing time fluctuation having linearity from the target chromatogram using the peaks detected on the reference chromatogram and the target chromatogram;
b) Based on the intensity of the peak detected in the target chromatogram after the linear correction by the linear correction means and the reference chromatogram, the peak is selected for each of the reference chromatogram and the target chromatogram, A coarse search means for searching for a candidate for a correspondence relationship between a reference chromatogram and a target chromatogram in a rough stage by performing matching by a dynamic programming algorithm for holding the peak retention time;
c) For the candidates for the correspondence relationship in the coarse stage extracted by the coarse search means, after adding the peaks excluded by the means, matching is performed by a dynamic programming algorithm for the peak retention time. A dense search means for searching for a correspondence relationship between the reference chromatogram and the target chromatogram in the dense stage,
d) correction processing means for correcting the time axis of the target chromatogram based on the correspondence between the reference chromatogram extracted by the dense search means and the target chromatogram;
A chromatogram data processing apparatus comprising:

Images