JPH02194379A - Base line distortion removing method for nuclear magnetic resonance - Google Patents

Abstract

PURPOSE:To remove the influence of the filter of a detection system and the influence of a transient response by a gate by irradiating a sample with a 90 deg. pulse and then with 180 deg. pulse before the start of sampling, and setting the time origin after the irradiation as a reference point. CONSTITUTION:When a 180 deg. spin echo pulse is inserted right before the sampling, the time origin can be shifted to a position where a taue (optionally set) time is elapsed after the pulse irradiation. Namely, if all spins slant to, for example, an X axis by the 90 deg. pulse irradiation, the spins spread on an XY plane at characteristic frequencies respectively thereafter, but when the sample is irradiated with the 180 deg. pulse and the phase is inverted by 180 deg. rotation around the X axis, all the spins are oriented to the X axis a specific time later. This timing is regarded as the time origin and then the time origin can be moved optionally considerably after the gate-on point of a filter system, so the time origin can be set after transient characteristic attenuation. Further, the influence of the DEAD time can be ignored because of the insertion of the spin echo.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method for removing baseline distortion in nuclear magnetic resonance (NMR).

[Conventional technology]

-1. In FT (Fourier transform) NMR, a high-frequency pulse is irradiated onto a sample, the response from the sample, that is, the free induction decay (FID) signal, is sampled, and the frequency spectrum of the response signal is obtained by Fourier transform. The elements contained in the samples are analyzed.

This kind of FTNMR takes in data over a finite interval and performs FT, but the baseline that is generated at that time changes due to a slight shift in the sampling start point, so it is not a problem as long as the baseline level is small. However, when it becomes large, it causes signal distortion.

This baseline distortion is not a big problem in normal one-dimensional NMR because the number of signals is not very large, but in two-dimensional NMR, noise that appears in the FTL, f, direction, so-called t, -noise Therefore, in two-dimensional NMR, it is required to remove baseline distortion after FT with particularly high accuracy.

Conventionally, as a baseline distortion removal method, 1
Regarding dimensional NMR, there is a method to find the baseline level for each section in the 0 frequency region and connect the levels for each section to make it the correct baseline ■If the frequency of the irradiation signal matches the resonance frequency, The Hoult method (JM
R51,110 ('83)) has been proposed.

However, methods ① and ② cannot be applied to two-dimensional NMR where there are a large number of signals.

In addition, regarding the two-dimensional NMR method, a device for the initial delay in the L3 stage has been proposed as a method to remove the baseline getter (t+-ridge) that runs in the fI axis direction (JM
R66,187 ('86)).

This means that, for example, if a monotonically attenuating signal S (t) as shown in FIG. 2(a) is obtained, 0. Δ,
If FT is performed by sampling 2Δ, 3Δ, etc., the shaded area in the figure will also be integrated as signal strength, and a baseline will be created by incorporating the shaded area. . Therefore, focusing on this point,
As shown in Figure 2(b), the sampling start point is set to Δ/
Shift by 2, Δ/2.3Δ/2.5Δ/2...
By sampling, the A part and the B part in the figure are canceled out to prevent the occurrence of a baseline.

[Problem to be solved by the invention]

However, even in the case of one-dimensional NMR, the above
, ■ It is impossible to completely remove baseline distortion.

Furthermore, the t, -ridge removal method described in (2) cannot be applied to real-time sampling in the L2 direction. This is because the detection system always includes a filter, which causes a time delay in the detected signal, and is also affected by the transient response after the gate is turned on to control signal detection, so simply start sampling by Δ/2. Delaying it doesn't work.

The present invention is intended to solve the above-mentioned problems, and is capable of eliminating the effects of filters in the detection system and the effects of transient responses caused by gates, thereby reliably removing baseline distortion in nuclear magnetic resonance. The purpose is to provide a removal method.

[Means to solve the problem]

To this end, the present invention irradiates a sample with a 90° pulse, receives a response signal, samples the response signal at predetermined time intervals, performs Fourier transform, and obtains a frequency spectrum. , a 180° pulse is irradiated before the start of sampling, and after a delay time due to the filter based on the time origin after the 180° pulse irradiation, 1/1 of the sampling interval is further emitted.
This method is characterized in that the signal is sampled after the time period 2 has elapsed.

[Effect]

In the present invention, a 180° pulse is irradiated after irradiation of 906 pulses and before the start of sampling, and after a time corresponding to the delay by the filter has elapsed based on the time origin after irradiation of the 180° pulse, 1/2 of the sampling interval is further irradiated. By sampling the signal after a certain period of time, baseline distortion can be removed by removing the influence of the delay and transient characteristics caused by the filter.

〔Example〕

Examples will be described below with reference to the drawings.

The baseline distortion removal method in the present invention consists of three elements: a correct FT method in digital FT (DFT), setting of an initial delay time in consideration of filter characteristics, and removal of filter transient characteristics. The explanation will be given below in order.

Now, let the signal time series S, be S, −e “”’ (j=o, 1.−) −−(+
) Interchange. -ω. +i" Δ: Sampling interval ω0: Resonance frequency ": Relaxation rate When the signal S is subjected to DFT, S (ω)=Σe Iα・ΔJΔ1−1 of
e''(--1tllA A.-ω 2.The first term of equation (2) is the spectrum to be obtained, and the second term is
The term Δ/2 is the gain of the baseline, and its integrated intensity is equal to the integrated intensity of the spectrum i/(:.-ω).

That is, A0-ω C0-ω+i "(ω.-ω)2+"2, and according to the sampling theorem, only the original signal can be extracted by integrating over the interval from -π/Δ to π/Δ.Therefore, When integrated over the real part, Δ 2 is obtained, and it can be seen that the integrated intensity of the baseline and the signal are equal.

(A) Therefore, since the offset that gives the baseline is the frequency 0 component, if you halve the size of the first sampling value that gives the frequency O component and perform normal sampling from the second, the effect of Δ/2 can be excluded.

(B) Also, as in the above-mentioned t, -ridge removal method (■), a delay time (Δ/2) is placed in the data acquisition, and -Δ
/2,3Δ/2,5Δ/2. ...Baseline distortion can be removed even if sampling is performed at (N+1)Δ/2 time.

However, in real-time sampling, a low-pass filter is always inserted into the signal system, and the results of the discussions in (A) and (B) above cannot be used unless the influence of this filter is correctly estimated and corrected.

The characteristics of the filter are broadly divided into (ii) changing the signal strength depending on the frequency; and (ii) changing the signal phase depending on the frequency. Commonly used filter characteristics include, for example, the Butterworth filter: H(ω)! -(1+ω0) Bessel filter is H(ω) ``IH(ω) l e-'''1RC1RC
Filter ω)-1/(1+iωτ) where τ is the time constant of the filter.

Now, when the signal e-h C6m'° is passed through a circuit as shown in Fig. 3 (this corresponds to the operation in the NMR receiving section), the output at this time is H(ωo) l e- ”””-”+ l H(ωo)
l f (tl) (4). Of course, this is the case when the Be5sel filter etc. works theoretically, but the response in the time domain is (4): (Effect of changing N signal strength...IH(ω.)(
II) Effect without shifting the time origin (quim shift)...
・・e ′w′lt・-tl (III) Effect of transient characteristics accompanying gate opening/closing...H(ω.) r (
t+') three types of effects appear. τ depends on the filter characteristics, and cannot be expressed in the cleanly separated form as shown above except for Be5sel filters. Among these, (II) and (1) are related to the baseline distortion of DFT.

The gist of the present invention is to correctly remove the time delay effect (n) and the transient response effect (III), and to perform the DFT method of the t, ridge removal method (■) explained in connection with FIG. .

FIG. 1 is a diagram showing a pulse immediately before data acquisition and a specific acquisition process.

First, an embodiment of the present invention for eliminating the time delay effect (II) will be described with reference to FIG. 1(a).

In FIG. 1(a), a delay time until the start of zumbling is set as determined by the following equation.

(90°ha JL/width) X 2 / π+DEAD
+DELAV=τ, +Δ/2 ・・・・・・・・・・・・
...(5) Here, in equation (5), the first term on the left side is a 90° pulse with an interpupillary distance of If you do, 9
It represents the elapsed time until the falling edge of the 0° pulse, and is a correction term for determining where the time origin is set. 2nd left side
The term DEAD is the dead time of the receiver, and the third term on the left side is DELAY.
is the delay from gate on to the start of sampling. Naturally, the shorter the DEAD time, the more desirable.

In equation (5), the first term on the right side is the delay time corresponding to the magnitude of the time origin shift caused by the filter, and in the case of the Be5sel filter, if fc is the filter cut frequency that provides a 3 dB reduction, then τ, = 2.1 /2πfc. The second term on the right side is the tl-
This is the time that gives the correct sampling start time for DFT using the ridge removal method, and is 1/1 of the sampling interval.
This is the delay time of 2.

In this way, after the time corresponding to the signal delay due to the filter inserted in the detection system has elapsed, by further shifting Δ/2 time and starting sampling, the j+-ridge
Can apply removal methods correctly.

Note that instead of DFT, continuous F T (confinu
It should be noted that the discussion held in ousFT is not fully applicable here (JMR51,010 ('83)).

In other words, if the FT of the continuous function x(t) is X(ω), and if there is a delay time τ, then the above-mentioned time delay (II) is simply a linear function proportional to the frequency. The spectrum can be corrected by phase correction. However, the tl-rid in FIG.
The results of the e-removal method (2) show that an incorrect sampling start causes not only a phase change on the spectrum but also a line distortion, and this idea cannot be applied. Therefore,
It is necessary to incorporate the correct time delay as shown in equation (5).

Next, an embodiment of the present invention that also eliminates transient characteristics will be described with reference to FIG. 1(b).

It goes without saying that the influence of transient characteristics can be removed by making the time constant of f (tl) in equation (4) very small, but generally speaking, this requires a very wide filter band and the signal contains high frequency components. Therefore, unnecessarily early sampling must be performed. In order to avoid this, in this embodiment, after irradiating 900 pulses, a 180° spin echo pulse is inserted immediately before sampling.

The spin echo pulse is effective because the time origin is not shifted within the 90° pulse but at a position τe after irradiation with the 1.80° pulse.

That is, by irradiating 90 pulses, all the spins are
If the spin falls on the axis, each spin will spread out on the X7 plane at a unique frequency, but after a predetermined time,
If a 0° pulse is irradiated and the phase is reversed by rotating 180° around the X axis, all spins will become
It will align with the axis. If this timing is shifted on the time origin, the time origin can be arbitrarily moved much later than the gate-on of the filter system, so it is possible to set the time origin after the transient characteristics have attenuated.

Therefore, as shown in FIG. 1(b), compared to FIG. 1(a), the order of the time origin at t=Q and the gate-on time is reversed. In this way, the transient response associated with gate-on can be attenuated before sampling starts. Furthermore, by inserting a spin echo, the influence of DBADtime can also be ignored. In the case of the above method, the sampling start is based on the time origin after the spin echo as the following equation.

Assuming the filter preset time is τ1, DELAY=τ, +Δ/2 DEAD-τe-τ1 ・・・・・・・・・・・・
...(6) (τe is set arbitrarily) As a specific experimental example, D E A D + D EL A
FIG. 4 shows a conventional example in which Y is misset and an example in which equation (5) or equation (6) is applied to two-dimensional NMR.

Figure 4(a) shows NOE S of oxytocin by conventional method.
FIG. 4(b) shows the N0Esy spectrum when the transient response according to the present invention is also removed.

It can be seen that the appearance of systematic noise running in the f2 direction differs between the case according to the present invention and the conventional case.

In addition, as a method for correcting the effect of the time origin used in the filter, there are also (a) oversampling to reduce the value of τ so as to obtain a spectrum width that is approximately twice the width of the frequency spectrum of the signal. How to eliminate distortion, (')fz
A method of subtracting systematic noise appearing in the force direction in the frequency domain is considered. Method (b) is a method that works although I don't know the reason, and (mouth) is t 1
- The exact same method as the correction on the frequency axis of f
This method is also applicable to systematic noise that appears in two directions, and does not work well when the noise is not entirely a direct current component but also has a slow frequency dependence.

〔Effect of the invention〕

As described above, according to the present invention, one-dimensional NMR22-dimensional NMR
By inserting 1800 spin-echo pulses after the last pulse in MR, we set DELAV to follow equation (6), completely removing the Heissline distortion, and further increasing the systematic speed running in the f2 direction in 2D NMR. It is possible to eliminate noise.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the pulse immediately before data acquisition and the specific acquisition process according to the present invention. FIG. 1(a) is a diagram showing an example of correcting signal delay, and FIG. Figure 2 shows an example of removing the t+-ridge.
Figure 3 is a diagram showing the detection system circuit, and Figure 4 (a) is a diagram explaining the removal method.
FIG. 4(b), which is a diagram showing the ESY spectrum, is a diagram showing the N0ESY spectrum according to the present invention, which also removes the influence of transient response.

Claims (1)

[Claims] (1) In Fourier transform NMR, a 90° pulse is irradiated to a sample, a response signal is received, and the response signal is sampled at a predetermined time interval and Fourier transformed to obtain a frequency spectrum. In NMR, sampling starts after irradiating a 90° pulse. Nuclear magnetism characterized in that a 180° pulse is first irradiated, and the signal is sampled after a delay time by a filter and after a time period of 1/2 of the sampling interval has elapsed based on the time origin after the 180° pulse irradiation. Baseline distortion removal method in resonance.