JP4486478B2 - Spectrum measurement method - Google Patents

Description

  The present invention relates to a spectrum measuring method for sampling a signal, subjecting this data to Fourier transform (FT) processing, and obtaining spectral information. The pulse NMR (nuclear magnetic resonance) method is an example to which the present invention can be applied.

  In general, a signal output from a measurement sample or the like is sampled, FT processing is performed on this data, and spectrum information is obtained to analyze the structure or the like of the measurement sample. Such a measurement technique is called a spectrum measurement technique.

  For example, in a pulse NMR method, which is a kind of spectrum measurement technique, spectrum information is obtained by performing FT processing on time axis data obtained by sampling an NMR signal from a measurement sample excited by an RF pulse. In the multidimensional pulse NMR method, multidimensional data including direct observation axes has spectrum information for performing FT processing.

  In spectral measurement methods such as pulse NMR, it is known that the resolution of the spectrum obtained is determined by the total sampling time, and the frequency domain of the spectrum is determined by the sampling interval (Nyquist theorem).

  The data sampling condition depends on the resolution and frequency domain of the observation target. When the resolution is high, the total sampling time is long, and when the frequency domain is wide, the sampling interval is short.

  Conventionally, in the multi-dimensional pulse NMR method, the measurement is performed with the sweep frequency region of the indirect observation axis being about half of the region to be originally obtained, and the peak outside the region is observed as a negative peak (referred to as the conventional first method). Has been used. At this time, when peaks outside the region are observed in the region, these peaks are called folded peaks.

  Conventionally, a method of performing sampling at random without performing sampling of NMR signals at equal intervals (referred to as a conventional second method) has been disclosed. In the case of this method, the obtained data corresponds to data obtained by randomly extracting data points from originally intended data. In order to infer the originally intended data from such data, this conventional method requires processing by the maximum entropy method.

  The maximum entropy method is a method for estimating data not including noise from data including noise. It is necessary to know in advance how much noise is included, and subjective processing is required. In addition, since a multi-stage process is required, a long time calculation using a computer is required.

  For example, the method described in the following non-patent document 1 will be described as an example. FIG. 14 is a diagram illustrating an example of spectrum reconstruction by the maximum entropy method of Non-Patent Document 1. According to this method, where data sampling is originally performed at 512 points, data sampling at 128 points determined at random is performed and analysis is performed so that entropy is maximized. Then, a spectrum capable of discriminating a sufficient peak as shown in FIG. 14 is obtained. FIG. 14A shows data sampling at 512 points. FIG. 14B shows a spectrum obtained by subjecting the 512-point sampling data to a Fourier transform process. FIG. 14C shows a spectrum obtained by spectrally reconstructing 128 sampling data determined at random by maximum entropy analysis.

  In this example, several minutes to several tens of calculations are required to reconstruct a one-dimensional spectrum, and several hours of calculation are required to process a multidimensional spectrum.

  In addition, it is necessary to manually specify the noise level in order to perform processing by the maximum entropy method.

Improved resolution in triple resonance spectra by non-linear sampling in the constant time domain (J. Bioml. NMR, 4, 483-490 (1994))

  By the way, in the first conventional method, whether a peak that should be outside the observation frequency region or a peak in the observation frequency region is determined based on whether it is a positive or negative peak. It was difficult to discriminate the peaks that were diffracted multiple times.

  In other words, since it is only possible to specify which region is originally doubled of the measured frequency region, the sweep frequency region of the indirect observation axis could not be reduced to about half. In addition, observation methods that can distinguish out-of-band peaks as negative peaks were limited.

  In addition, since the conventional second method of performing random sampling and reconstructing free induction attenuation (FID) by the maximum entropy method cannot perform fast Fourier transform, it consists of a large number of sampling points such as a multidimensional NMR method. As described above, the data processing requires a very long computer processing. Further, as described above, it is necessary to manually specify the noise level in order to perform processing by the maximum entropy method.

  The present invention has been made in view of the above circumstances, and an object thereof is to provide a spectrum measurement method capable of acquiring a spectrum in a short time with a small amount of sampling data.

  In order to solve the above-described problems, the spectrum measurement method according to the present invention performs narrow-band Fourier transform at a plurality of different sampling rates, and shifts the center frequency of the spectrum by the observation frequency range until the target broadband is obtained. And a logical product calculating step for calculating a logical product of the plurality of types of the stacked spectra.

  In the stacking step, it is preferable that the whole spectrum is reconstructed in consideration of aliasing, and in the logical product calculation step, signals whose positions coincide with each other in the spectra are selected to obtain the original spectrum.

  The spectrum measurement method of the present invention is preferably applied to one-dimensional NMR measurement or multi-dimensional NMR measurement.

  In order to solve the above problems, the spectrum measurement method according to the present invention performs a Fourier transform process on a plurality of types of FIDs in a narrow band, and sets the center frequency of the spectrum only in the observation frequency range until the target wide band is obtained. A step of repeating the process of stacking with a shift, and a logical product calculation step of calculating a logical product of the plurality of types of stacked spectra.

  In general, data sampling is performed at a sampling interval determined by a frequency region to be observed. The spectrum measurement method of the present invention reduces the total measurement data by observing a spectrum under the same conditions over a wide range of sampling intervals, and performs measurements at a plurality of sampling intervals. As a result, spectrum information having a normal frequency range to be observed can be obtained. Since the total measurement data can be reduced, the total measurement time can be shortened.

  In other words, the spectrum measurement method of the present invention performs narrowband NMR measurement at a plurality of different sampling rates, then reconstructs the entire spectrum in consideration of aliasing, and signals whose positions coincide with each other. To select the original spectrum. Thereby, an NMR spectrum is acquired in a short time with a small amount of data.

  Specifically, in the present invention, by performing measurement using two or more types of sweep frequency regions of different indirect observation axes, it is possible to determine the folding of a peak from a region more than twice the measured frequency region. As a result, the conventional method has shortened the distance between the measurement temples by about half, but it has become possible to further shorten it.

  According to the spectrum measurement method of the present invention, since the number of sampling data points is small, the capacity of the data storage memory can be reduced. In addition, when applied to the indirect observation axis of multidimensional measurement, a time shortening effect according to the number of sampling data points can be obtained. In addition, high-dimensional NMR measurement, which requires a very long measurement time and is actually impossible to measure, can be performed with a measurement time that can be actually measured. In addition, when the measurement is performed with the same measurement time as before (when the number of sampling data points is reduced and the number of integrations of each scan is increased), the sensitivity is increased due to the integration effect. Furthermore, wide-band measurement is possible with short-time narrow-band measurement.

  Hereinafter, the best mode for carrying out the present invention will be described. FIG. 1 is a block diagram of a pulse NMR measurement apparatus 1 or 1 ′ used in each of Examples 1 to 4 below. According to each example, the pulse NMR measurement apparatus performs one-dimensional NMR measurement and multidimensional NMR measurement.

Example 1
The pulsed NMR signal obtained with a pulsed NMR measurement device is free induction decay (FID). Example 1 is a pulse NMR measurement apparatus that performs a spectrum measurement method for sampling a pulse NMR signal FID and further performing a Fourier transform (FT) process.

  In FIG. 1, the pulse NMR measuring apparatus 1 employs a configuration including a superconducting magnet unit 10, a spectrometer unit 20, and a computer 50.

  The spectrometer unit 20 of the pulse NMR measuring apparatus 1 performs digitized phase detection on a signal detected by an output system 30 that transmits one type of frequency and outputs it to the magnet unit 10 and a probe of the magnet unit 10. And a detection system 40 for detecting FID.

  The output system 30 includes a frequency synthesizer 31 that transmits one type of frequency, a transmitter 32 that performs pulse shaping by shaping the output of the frequency synthesizer 31, and a power amplifier 33 that amplifies the transmitter output.

  The detection system 40 includes an amplifier 41 that amplifies the probe detection signal, a detection unit 42 that performs FID detection from the amplified probe detection signal, and an A / D converter 43 that digitizes the NMR signal FID detection output.

The magnet unit 10 is composed of a superconducting coil made of NbTi and Nb ₃ Sn, for example, in order to set the measurement magnetic field to 600 MHz, for example. In the magnet section, a weak signal from the sample is detected, and a probe 11 serving as a transmission / reception coil for sending a pulsed electromagnetic wave to the sample, and a preamplifier 12 that amplifies the signal detected by the probe 11 and supplies the signal to the detection system. Is included.

  The computer 50 has a central processing unit (CPU), and a ROM, RAM, HDD, and I / O interface are connected to the CPU via an internal bus. An input device such as a keyboard and a mouse and an output device such as a display unit and an audio output unit are connected to the I / O interface. The computer 50 is connected to the output system 30 and the detection system 40 of the spectrometer unit 20 via a communication interface, controls the frequency of the frequency synthesizer 31, and receives digital detection data from the A / D converter 43. Data processing. Specifically, the sampling data from the A / D converter 43 is subjected to Fourier transform (FT) processing to detect a spectrum.

  The spectrum measurement method according to the present invention is executed by the computer 50 as an NMR spectrum measurement program. The NMR spectrum measurement program is stored in, for example, a ROM or HDD, read out to the RAM by the CPU, and executed using the RAM as a work area. The computer 50 executes the NMR spectrum measurement program, thereby controlling the sampling process of the A / D converter 43 of the detection system 40 and applying a Fourier transform process to the data sampled by its own CPU.

  FIG. 2 is a diagram showing a signal processing relationship between the A / D converter 43 and the computer 50. By executing the NMR spectrum measurement program, the Fourier transform unit 51 functions in the computer 50.

  Details of the operation will be described below. The pulse NMR signal FID detected by the detection unit 42 in the detection system 40 of the pulse NMR measurement apparatus 1 is supplied to the A / D converter 43. At this time, when the length of the target pulse NMR signal FID is 1 s and the data point (sampling) interval is 1 ms, the total number of data points is 1000. In this case, in order to sample all the data points, it is necessary to measure 1000 points.

  FIG. 3 is a diagram showing a relationship between a sampling point and an observation frequency band in NMR. FIG. 3A shows an example of sampling points when a wide frequency range is observed. FIG. 3B shows an example of sampling points when a narrow-band frequency region is observed. Since the observed FID length is the same for both (a) and (b), the resolution does not change, but the observation frequency range is different because the sampling point interval is different. When observed in a narrow band, peaks outside the observation frequency range are observed as folded peaks.

  The frequency band of the spectrum obtained by Fourier transforming this data is 1 kHz (reciprocal of the sampling interval), and the resolution is 1 Hz (frequency region / number of data points).

  Considering the case where the sampling interval is set to 5 ms while maintaining the resolution, the number of data points is 200 when the FID having a length of 1 s is observed at intervals of 5 ms. In this case, the sampling points are 5 ms, 10 ms, 15 ms, 20 ms,... 1000 ms.

  When this data is Fourier transformed, the resolution is the same as when 1000 points are sampled, but the frequency band becomes 200 Hz, and the peak outside this band appears in the spectrum as a folded peak.

  For example, when measuring in the observation frequency range from 1000 Hz to 2000 Hz, the same measurement is performed with the observation frequency range changed from 1000 Hz to 1200 Hz for the samples that have peaks at 1300 Hz, 1420 Hz, and 1640 Hz. In normal quadrature detection, peaks are observed at 1100 Hz, 1020 Hz, and 1040 Hz.

  FIG. 4 shows a measurement example in a narrow band. (A) is a measurement example at 200 Hz, and (b) is a measurement example at 250 Hz. In the example of Fig. 4 (a), the peak from 1000Hz to 1200Hz is not folded, the peak from 1200Hz to 1400Hz is folded once from 1000Hz to 1200Hz, and the peak from 1400Hz to 1600Hz is from 1000Hz to 1200Hz. The peak from 1600 Hz to 1800 Hz is observed as the third diffraction peak from 1000 Hz to 1200 Hz.

  For the peak observed in this way, the number of turns cannot be obtained from this result.

  Therefore, when another measurement (FIG. 4B) is performed in a band from 1000 Hz to 1250 Hz, in the case of the peak in this example, peaks are observed at positions of 1050 Hz, 1170 Hz, and 1140 Hz. In this case, the sampling points are 4 ms, 8 ms, 12 ms, 16 ms, 20 ms,... 1000 ms.

  Here, the peak observed at the 1100 Hz position in the band 200 Hz measurement is known to be observed from the original position from 1000 Hz to 2000 Hz, and in consideration of aliasing, the correct position candidates are 1100 Hz, 1300 Hz, 1500Hz, 1700Hz and 1900Hz are considered.

  On the other hand, when the peak observed at the position of 1050 Hz is considered in the measurement of the bandwidth of 250 Hz, 1050 Hz, 1300 Hz, 1550 Hz, and 1800 Hz are considered as candidates for the correct position in the range of 1000 Hz to 2000 Hz in consideration of aliasing. Comparing these candidates and finding the frequency common to both by AND processing (logical product), it can be seen that the correct position is 1300 Hz.

  Similarly, 1420Hz is the correct position for the peaks observed at 1020Hz and 1170Hz in the band 200Hz and 250Hz measurements, and 1640Hz is the correct position for the peaks observed at 1040Hz and 1140Hz.

  This processing can be performed according to the processing procedure shown in FIG. First, measurement A in a narrow band is performed (step S1). A normal Fourier transform process is performed (step S2), and the process of stacking (turning back) the center frequency of the spectrum by shifting it by the observation frequency range is repeated until the target wide band is obtained (step S3). Similarly, measurement B in a narrow band is performed (step S4), a normal Fourier transform process is performed (step S5), and the center frequency of the spectrum is shifted by the observation frequency range and stacked until the target wide band is obtained. Repeat (step S6). Then, an AND process is performed on the spectrum obtained by measurement A and the spectrum obtained by measurement B (step S7), and common peaks are extracted.

  A specific example of the processing procedure will be described with reference to FIGS. FIG. 6 is a plot example in which spectra measured in a narrow band are stacked. FIG. 6A shows an example in which spectra measured at 200 Hz are stacked. FIG. 6B shows an example in which spectra measured at 250 Hz are stacked. FIG. 7 is a diagram showing a result of performing an AND process on FIGS. 6A and 6B.

  First, the spectrum measured in the band 200Hz is a spectrum from 1000Hz to 1200Hz, 1200Hz to 1400Hz, 1400Hz to 1600Hz, 1600Hz to 1800Hz, 1800Hz to 2000Hz. (FIG. 6A).

  Next, the spectrum measured in a band of 250 Hz is a spectrum from 1000 Hz to 1250 Hz, from 1250 Hz to 1500 Hz, from 1500 Hz to 1750 Hz, and from 1750 Hz to 2000 Hz, and the connected spectrum is a spectrum from 1000 Hz to 2000 Hz (FIG. 6B). ).

  Next, the spectrum obtained by the above two processes is ANDed. That is, the two spectra obtained by the two processes are compared, and a common peak is extracted as shown in FIG.

  The spectrum obtained by the above processing procedure (Fig. 7) is the same spectrum as the spectrum measured in the 1000 Hz to 2000 Hz band, with the same resolution and band, and the number of measurement points is 45% (band 200 Hz measurement: 20% + band 250 Hz). Measurement: reduced to 25%).

  As described above, the present invention configures a plurality of existing peaks into an original spectrum at a correct frequency position from the stacked peak spectrum under the condition that the spectrum is sufficiently separated so that the peak does not have a large width. It is a method to do. With this method, the number of measurement points can be reduced. However, a ghost peak may remain due to the relationship between the difference in frequency between peaks and the difference in frequency band between two types of spectra.

  In such a case, the reliability can be improved by changing the acquired spectrum from two types to three types.

  In this example, the spectrum has peaks observed at 1300Hz, 1420Hz, and 1640Hz. If the difference in frequency between the peaks is equal to the difference in frequency band between the two types of spectrum, this procedure can be performed as a ghost peak. A peak remains in the wrong position.

  Change the spectrum to be acquired from 2 types to 3 types. For example, if you measure a spectrum with peaks observed at 1300 Hz and 1350 Hz under the above conditions, the peaks will appear at 100 Hz and 150 Hz when the sweep frequency band is 200 Hz. Observed, when measured at 250 Hz, peaks are observed at 50 Hz and 100 Hz. At this time, in order to determine at which position the peak observed at 100 Hz is to be observed, a third measurement can be performed by setting the sweep frequency band to 180 Hz or the like. When measured at 180 Hz, peaks are observed at 120 Hz and 170 Hz.

Example 2
This Example 2 is also a pulse NMR measurement device that samples a pulse NMR signal FID and further executes a spectrum measurement method for performing a Fourier transform (FT) process, but the example applied to the two-dimensional NMR measurement is Example 1. And different.

  In FIG. 1, the pulse measurement device 1 ′ is different from the configuration of the first embodiment in that the pulse measurement device 1 ′ includes an output system 30 ′ that simultaneously transmits two or more types of frequencies and outputs the same to the magnet unit 10. Since the other configuration is as described above, the description is omitted here.

  The operation will be described below. In the case of Example 1, the FID was obtained by changing the conditions of the direct observation axis of the one-dimensional NMR measurement. However, the number of measurement points can be reduced and the measurement time can be shortened even in the multi-dimensional measurement that samples the indirect observation axis. .

FIG. 8 shows an example of two-dimensional NMR measurement in which the observation frequency band of the indirect observation axis of ¹⁵ N- ¹ H HSQC of ¹³ C / ¹⁵ N-labeled human ubiquitin is narrow. FIG. 8A shows a spectrum measured in the frequency band 1 kHz (20 points) of the indirect observation axis, and FIG. 8B shows a spectrum measured in the frequency band 1.1 kHz (22 points) of the indirect observation axis.

  As can be seen from FIG. 8, the spectrum measured in a narrow band is very difficult to analyze because the folded peak is observed in a narrow range.

  FIG. 9 shows an example in which two-dimensional HSQC spectra measured in a narrow band are stacked. 9A shows a spectrum measured in the frequency band 1 kHz of the indirect observation axis, and FIG. 9B shows a spectrum measured in the frequency band 1.1 kHz of the indirect observation axis. The spectrum is connected in the same manner as in Example 1. As can be seen from FIG. 9, all the compensation for the position where the peak should be originally observed is included.

  In this example, since the quadrature detection of the indirect observation axis is performed by STATES-TPPI, the peak is folded in the same way as the quadrature detection of the direct observation axis. Depending on the method of quadrature detection, the pattern of the odd diffraction return peak may be reversed.

  When FIG. 9A and FIG. 9B are compared, the peak that appears in (a) may not be seen in (b) because the conditions under which the peak turns back are different.

  That is, the peak that should be observed originally can be seen in both FIG. 9A and FIG. 9B, but the peak that appears only in either of FIG. 9A or FIG. This means a certain folded peak, and by extracting a common peak by AND processing, it is possible to obtain only the originally intended peak.

  10 to 12 show the results of Example 2 (including a conventional comparative example). FIG. 10A shows a spectrum obtained by reconstructing the frequency range of the indirect observation axis from spectra of 1 kHz and 1.1 kHz. The measurement time was about 35 minutes. FIG. 10B shows a spectrum obtained by reconstructing the frequency range of the indirect observation axis from the spectra of 1 kHz, 1.1 kHz, and 1.2 kHz. The measurement time was about 56 minutes. FIG. 10C shows a spectrum measured with the frequency range of the indirect observation axis set to 5 kHz. The measurement time was about 85 minutes.

FIG. 11A shows a spectrum obtained by reconstructing the frequency range of the indirect observation axis from spectra of 1 kHz and 1.1 kHz. The measurement time was about 35 minutes. 10 with respect to (a), the horizontal axis ^{9.5~7.7 (1 H [ppm])} × vertical axis 129.0~113.0 ⁽¹⁵ N [ppm]) is an enlarged view of. FIG. 11B shows a spectrum obtained by measuring the frequency range of the indirect observation axis at 5 kHz. The measurement time was about 85 minutes. 10 with respect to (c), the horizontal axis ^{9.5~7.7 (1 H [ppm])} × vertical axis 129.0~113.0 ⁽¹⁵ N [ppm]) is an enlarged view of.

FIG. 12A shows a spectrum obtained by reconstructing the frequency range of the indirect observation axis from spectra of 1 kHz, 1.1 kHz, and 1.2 kHz. The measurement time was about 56 minutes. 10 with respect to (b), the vertical axis is a diagram was ^{140~90.0 (15 N [ppm])} . FIG. 12B shows a spectrum obtained by measuring the frequency range of the indirect observation axis at 5 kHz. The measurement time was about 85 minutes. 10 with respect to (c), the vertical axis is a diagram was ^{140~90.0 (15 N [ppm])} .

  The spectrum reconstructed from FIGS. 8 (a) and 8 (b) in the same procedure (FIG. 5) as in Example 1 is shown in FIG. 10 (a), and FIGS. 11 (a) and 11 (b). As can be seen from the comparison of), it can be seen that the same resolution as the spectrum observed in a normal broadband experiment (5 kHz / 100 points) is obtained.

  In FIG. 10 (a), a ghost peak is obtained on the same axis where a plurality of peaks are arranged vertically. As a third measurement, a spectrum measured in a frequency band of 1.2 kHz (24 points) is used. By performing AND processing on the three types of stacked spectra, a more reliable spectrum can be obtained as shown in FIG.

  Furthermore, as can be seen from FIG. 12, the frequency range of the spectrum reconstructed from the spectra measured at 1 kHz, 1.1 kHz, and 1.2 kHz is 5 kHz or more (if there is only one peak, it is the best for analysis. Least common multiple of a plurality of frequency bands), a peak observed as a folded peak in FIG. 12B also appears in the observation frequency range.

  Note that since the observation range in FIG. 12B is from 91.4 ppm to 144.7 ppm, the peak near 86 ppm is observed as a peak folded around 138 ppm.

  In this case, 100 points are sampled as indirect observation axes in FIG. 10 (c), whereas 42 points are sampled in FIG. 10 (b) and 66 points are sampled in FIG. 10 (a). Therefore, as the number of measurement points is reduced, the measurement time is shortened to 42% and 66%, respectively.

Example 3
This Example 3 is also applied to multidimensional NMR measurement. This is a pulse NMR apparatus 1 ′ capable of multidimensional measurement shown in FIG. The configuration of the apparatus is omitted.

  The operation will be described below. The processing procedure (FIG. 5) performed in Examples 1 and 2 is effective for all indirect observation axes of multidimensional NMR measurement.

  Therefore, in the three-dimensional XYZ coordinate system, the observed frequency band of the Y-axis and Z-axis is narrowed and the spectrum obtained by the three-dimensional measurement is stacked in the Y-axis and Z-axis direction, and in two (or more) different frequency bands. By extracting a common peak from the observed spectrum, the same effect as in Examples 1 and 2 can be obtained.

  In this case, the total measurement time reduction effect is the product of the Y-axis time reduction effect and the Z-axis time reduction effect, and the measurement can be performed by reducing the Y-axis by 66% and the Z-axis by 66%. If done, the total measurement time can be reduced to the product of 66% and 66%, or 44%.

  Next, it is possible to shorten the time for measurements of four or more dimensions by the same method, and when measuring in a frequency band that reduces the measurement time of the Y-axis, Z-axis, and A-axis by 66% in the four-dimensional measurement, Total measurement time is reduced to the third power of 66% or 29%.

  Since the frequency position of the peak observed on the same axis of multi-dimensional NMR measurement exceeding two dimensions can be predicted by performing two-dimensional NMR measurement as a preliminary experiment, in the multi-dimensional NMR experiment, spectra of two types of frequency bands are used. It is enough to observe.

  Furthermore, in such a high-dimensional measurement, the frequency at which a plurality of peaks are observed on the same axis is very low. Therefore, unlike in the case of two-dimensional measurement, in most cases, only measurements with two different conditions are sufficient. Therefore, it can be expected that the higher the measurement, the smaller the number of measurement points and the shorter the measurement time.

Example 4
Example 4 is a multi-dimensional NMR pulse measuring apparatus that performs measurement by multi-linear sampling. 2 is a pulse NMR apparatus capable of multidimensional measurement with the configuration of FIG.

  The operation will be described. In Examples 1, 2, and 3, measurement is repeated under a plurality of different measurement conditions, but data sampling at the same timing may be performed even in measurements in different frequency bands.

  For example, when data sampling at intervals of 4 ms and data sampling at intervals of 5 ms are performed, data sampling points corresponding to multiples of 2 ms, 40 ms,.

  In such a case, the data sampling point is 4 ms, 5 ms, 8 ms, 10 ms, 12 ms, 15 ms, 16 ms, 20 ms, etc., and data sampling is performed at a timing that is a multiple of 4 ms and 5 ms. By extracting points that are multiples of 5 ms, the problem of sampling common sampling points can be avoided.

  Moreover, when measured at 1 kHz, 1.1 kHz, and 1.2 kHz in Example 2, three types of measurement results can be extracted from 62 points, and the measurement time can be shortened from 66% to 62%.

  Sampling at a multiple of two or more intervals corresponds to the fact that a plurality of linear samplings (multi-linear sampling) are performed simultaneously.

  The plurality of FIDs extracted in this way are equivalent to the FIDs obtained by performing a plurality of measurements in Examples 1, 2, and 3. Therefore, the broadband spectrum is reproduced by the same procedure as in Examples 1, 2, and 3. Configuration is possible.

  FIG. 13 is a flowchart illustrating a measurement processing procedure in multi-sampling including sampling intervals A and B performed in the fourth embodiment. First, FIDs corresponding to the narrow band A and the narrow band B are extracted (step S11).

  A normal Fourier transform process is performed on the narrow band FID A (steps S12 to S13), and the process of stacking the spectrum by shifting the center frequency of the spectrum by the observation frequency range is repeated until the target broadband is obtained (step S14). . Similarly, the normal Fourier transform process is performed on the narrow band FID B (steps S15 to S16), and the process of stacking the spectrum with the center frequency shifted by the observation frequency range is repeated until the target wideband is obtained ( Step S17). Then, AND processing is performed on the spectrum obtained by narrowband FIDA and the spectrum obtained by narrowband FID B (step S18), and common peaks are extracted.

It is a block diagram of a pulse NMR measuring device. It is a figure which shows the signal processing relationship of an A / D converter and a computer. It is a figure which shows the relationship between the sampling point in NMR, and an observation frequency band. It is a figure which shows the example of a measurement in a narrow band, and shows the example of a measurement in 200Hz, and the example of a measurement in 250Hz. 3 is a flowchart illustrating a processing procedure in the first embodiment. It is a characteristic view which shows the example of a plot which piled up the spectrum measured by the narrow band. FIG. 7 is a characteristic diagram showing an AND processing result of the spectrum measured at 200 Hz and the spectrum measured at 250 Hz shown in FIG. 6. The observation frequency band of an indirect observation axis in ¹⁵ N- ¹ H HSQC is a diagram illustrating an example spectrum of a narrow band two-dimensional NMR measurements were measured. It is a figure which shows the example which piled up the two-dimensional HSQC spectrum measured in the narrow band based on the spectrum measuring method by this invention. It is a figure which shows the measurement result by Example 2. FIG. It is a figure which shows the measurement result (enlargement) by Example 2. It is a figure which shows the measurement result (vertical axis extension) by Example 2. FIG. 10 is a flowchart illustrating a measurement processing procedure in multi-sampling performed in Example 4; It is a figure which shows the example of the spectrum reconstruction by the maximum entropy method of a nonpatent literature 1.

Explanation of symbols

S1 A step of performing measurement A in a narrow band, S2 a step of performing normal Fourier transform processing, S3 a step of repeating a process of shifting and stacking the center frequency of the spectrum by the observation frequency range until the target wide band is obtained, S4 narrow A step of performing measurement B in the band, S5 a step of performing a normal Fourier transform process, S6 a step of repeating the process of stacking the spectrum with the center frequency shifted by the observation frequency range until the desired wideband is obtained, S7 obtained by measurement A Performing AND processing on the obtained spectrum and the spectrum obtained by measurement B

Claims (4)

Performing a narrowband Fourier transform at multiple different sampling speeds, shifting the center frequency of the spectrum by the observation frequency range and stacking up to the target wideband,
And a logical product calculating step of calculating a logical product of the plurality of stacked spectra.   2. The stacking step reconstructs an entire spectrum in consideration of aliasing, and the logical product calculation step selects a signal having a matching position between the spectra to obtain an original spectrum. The spectrum measuring method as described.   The spectrum measuring method according to claim 1, which is applied to one-dimensional NMR measurement or multi-dimensional NMR measurement. Applying a Fourier transform process to a plurality of types of FIDs in a narrow band and repeating a process of stacking the center frequency of the spectrum by shifting it by the observation frequency range until the target wide band is obtained;
And a logical product calculating step of calculating a logical product of the plurality of stacked spectra.

Images