DE3914302C2 - Process for recording spin resonance spectra - Google Patents

Description

The invention relates to a method for absorbing spin resonance spectra of samples with at least two groups of Cores of the same nucleus, whose chemical shift in the essentially coincides, for sole illustration  the signal of the first group the signal of the second group is suppressed.

In the technique of spin resonance spectroscopy it is known Spectra where signals from different core groups are at each other overlay to "edit". This means different Recording techniques with which it is possible to overlay Filter out spectra of individual signals. Usually happens This is done by taking several measurements with different ones Measurement parameters carried out and then unwanted signal components be eliminated by difference formation.

Examples of such editing techniques for nuclear magnetic resonance applications are in the textbook by Sanders, Jeremy K.M. and Brian K. Hunter "Modern NMR Spectroscopy", Oxford University Press, 1987, pages 237 to 259. Other such ver Driving are in EP-OS 2 44 752 and EP-OS 1 66 559 be wrote. In these other known methods uncoupled spins by forming the difference between two measurements suppressed.

However, the editing techniques described above have the common disadvantage that to record a spectrum several, staggered measurements with different Measurement parameters must be carried out. While this is at long-lasting chemical samples in a laboratory are not essential Represents a problem, there may be significant problems if such nuclear magnetic resonance spectra on biological samples, namely to be performed on living tissue. This applies especially when in vivo measurements on patients be carried out and movement artifacts related to measured value falsification can lead.  

Differential measuring methods also have the basic one Disadvantage that due to the subtraction of high interference signal Amplitude measurement errors can occur in the same sizes order lie like the useful signal.

It is also known to selectively select nuclear magnetic resonance spectra, i.e. only for a certain, spatially defined area to take a sample. This recording technique has special Importance in biological research as well as medicine. With this recording technique it is possible, for example have a nuclear magnetic resonance spectrum of a patient certain defined point in an internal organ of the Admit patients.

The acquisition of volume selective nuclear magnetic resonance spectra is a technique known per se, examples can be found in the textbook by Wehrli, Felix W., Derek Shaw and J. Bruce Kneeland "Bio medical Magnetic Resonance Imaging ", Verlag Chemie, 1988, Pages 1 to 45 and 521 to 545.

From the US-Z "Journal of Magnetic Resonance" 67 (1986), Page 148, is e.g. the so-called SPARS process and from the US-Z "Journal of Magnetic Resonance", 68 (1986), page 367, the so-called DIGGER process is known. With these, The processes are volume-selective processes, where the layers are outside the selected volume area are saturated, so that only the selected Volume area remains. A disadvantage of these two known processes, in particular in the DIGGER process, that a high radio frequency power is needed, and at in both of these methods, the presaturation  High-frequency pulses can be set very precisely, because otherwise additional signals are generated.

Another special procedure for volume selective recording of nuclear magnetic resonance spectra using three in time spaced 90 ° high-frequency pulses with simultaneous Apply different gradient magnetic field pulses in below Different coordinate directions is also in DE-OS 34 45 689. In this known method generates conventional stimulated spin echoes.

From US-Z "Magnetic Resonance in Medicine", 9 (1989), Pages 254 to 260, it is finally known to be volume selective Spectra using homonuclear polarization edit transfers using a one-time transfer becomes.

If the nuclear magnetic resonance spectrum is to be recorded on a homonuclear or heteronuclear coupled spin system of the general form A _n X _m , as is of great interest for biomedical research, because such measurements allow conclusions to be drawn about the metabolism in organic tissue, it often arises the problem of signal overlap. In the case of a homonuclearly coupled spin system, both coupling partners consist of the same nucleus, e.g. of protons ( ¹ H), while in the case of heteronuclearly coupled spin systems the coupling partners belong to different nucleus types, e.g. the A group to the protons ( ¹ H) and the X group a carbon isotope ( ¹³ C). In a homonuclear example of lactate, an A₃X system, for example the methyl group (CH₃) has approximately the same chemical shift, ie the position of the line in the spectrum as the CH₂ group of lipid, because both chemical shifts are approximately at 1.35 ppm lie. However, since the lipid concentration in living tissue can be substantially larger, the CH₃ signal of the lipid covers the CH₃ signal of the lactate. The same applies in general to the editing of heteronuclear A _n X _m systems, for example for an A₃X system such as methanol with ¹³ C enrichment.

Would you now be a volume selek in the former example tive lactate measurement in a lipid environment with the known Make editing techniques where, as mentioned, two Measurements in chronological order with different parameters problems would arise, if the patient is moving during these two measurements. This is because artefacts arise from the two measurements are influenced differently, so that connect to one the difference formation not only the desired isolated CH₃ signals of lactate, but also the unwanted lipid Artifacts can be prepared by editing.

Although the invention is within the scope of the present application based on an application in the field of nuclear magnetic resonance (NMR) is explained, it goes without saying that it also applies to other forms the spin resonance can be used, especially with paramag netic electron resonance (EPR) or nuclear / electron double resonance techniques (ENDOR, ELDOR, NEDOR, Overhauser etc.).

The invention is further illustrated by the simple example of the scalar Coupling (J) shown, but it is understood that Anwen with other types of coupling, e.g. the dipole coupling, possible are.  

It is also for selective editing of metabolite signals known, multi-quantum transitions (MQ) or zero quantum transitions (ZQ). Examples of this can be found in US-Z "Journal of Magnetic Resonance", 64 (1985), page 38, in the US-Z "Journal of Magnetic Resonance" 78 (1988), page 355, in US-Z "Journal of Magnetic Resonance", 77 (1988), Page 382, and the US-Z "Magnetic Resonance in Medicine", 9 (1989), page 32.

The main disadvantage of difference-forming methods lies in their dependence on precise subtraction. This is an essential aspect in in vivo measurements, at where dominant resonances of water and lipid occur can. Another disadvantage is the need for two dimensional measurements to extra the metabolite signal here. In a known one-dimensional process selective excitation and pulsed field gradients are used, to select only zero quantum coherences so that good spectral editing with a single measurement suppression of solvents was possible. Indeed these methods have the major disadvantage of being a loss signal-to-noise ratio occurs because always both zero quantum and multi-quantum coherences are excited, but only one of these coherences into one observable A quantum signal (SQ) is reshaped while doing that others dephased by a field gradient or by Um switching the phase is eliminated. If zero quantum coherences only a quarter of the coherence is selected for a subsequent application of field gradients to ver addition, and the effectiveness of a desired training of Multi-quantum coherences can be derived from the large number of spin-spin  Couplings depend, for example in the case of ethanol the one-quantum and three-quantum fractions (SQ, TQ) are lost go. It has also been suggested selective polarisa tion transfer procedures, but also lead them loss of signal due to the direction of the transfer or due to the dependence of the polarization transfer on High frequency pulse angle and therefore require extensive cyclical changeover of the phase. More about the latter Effects is in US-Z "Journal of Magnetic Resonance", 66 (1986), page 86, and in the US-Z "Progress in NMR Spectro scopy "16 (1983), page 163.

In contrast, the invention is based on the object To further develop methods of the type mentioned at the beginning, that editing desired spectral components with only one Measurement is possible without losing signal amplitude.

According to the invention, this object is achieved by

• - that a pulse train of four high-frequency pulses, preferably 90 ° high-frequency pulses to which Sample is irradiated,
• - That the first high-frequency pulse one for the Cores of the first group of selective, soft impulse is
• - That the second high-frequency pulse one for the Cores of the second group of selective, soft impulse is and is set such that the Magneti the cores of the first group and the second group at least partially in one state of multi-quantum coherence,  
• - that immediately after the second high frequency pulse the third high-frequency pulse as at least one hard impulse to the test irradiated and adjusted so that the Multi-quantum coherence in a state of correlier ten-order is transferred,
• - That in the time interval between the third high frequency pulse and the fourth high-frequency pulse a dephasing gradient magnetic field pulse is applied to the test, and
• - That the fourth high-frequency pulse for the Cores of the first group of selective, soft impulse is and is set so that the More quantum coherence back into a state of input Quantum coherence of the nuclei of the first group over leads, and
• - that then the resonance signal of the cores of the first Group is detected.

The object underlying the invention is also thereby solved,

• - that a pulse train of four high-frequency pulses, preferably 90 ° high-frequency pulses to which Sample is irradiated,
• - That the first high-frequency pulse one for the Cores of the first group of selective, soft impulse is
• - That the second high-frequency pulse one for the Cores of the second group of selective, soft impulse  is and is set such that the Magneti the cores of the first group and the second group at least partially in one state of multi-quantum coherence,
• - that immediately after the second high frequency pulse the third high-frequency pulse as at least one hard impulse to the test irradiated and adjusted so that the Multi-quantum coherence in a state of correlier ten-order is transferred,
• - That in the time interval between the third high frequency pulse and the fourth high-frequency pulse a dephasing gradient magnetic field pulse is applied to the test, and
• - That the fourth high-frequency pulse for the Cores of the second group of selective, soft impulse is and is set such that the state the correlated z-order using polarization transfer from the cores of the first group to one State of single-quantum coherence of the nuclei of the second group is transferred, and
• - that then the resonance signal of the cores of the first Group is detected.

The method according to the invention uses a so-called ECZOTIC Pulse sequence (editing via Correlated Z-Order with Total Inherent Coherence), where all metabolite coherences are selective into a correlated Z-order or a longitudinal two- Spin order can be converted, similar to known methods, as they became known under the keyword INEPT,  see. e.g. US Progress in NMR Spectroscopy, 16 (1983), Page 163.

Here, "z-order" means that before applying a magnet field gradients the magnetizations correlate in the z-direction (Field direction) are brought so that they are in this gradient do not dephase; unlike the rest of the distracting Coherences.

Dephasing thus follows the conversion mentioned the other signals by means of pulsed field gradients, so that the mentioned orders into observable single-quantum coherents zen can be converted with maximum signal yield. The Insensitivity of zero quantum coherences to one Field inhomogeneity is known and already the basis of editing methods that is correlated z-order however, just like any longitudinal magnetization, also invariant in the presence of a field gradient and therefore offers an alternative approach to a process for spectral editing. The method according to the invention can basically applied to any spin multiplet, in the context of the present application, however, the spectral Editing of lactate and ethanol described in detail become.

In a particularly preferred embodiment of the invention According to the method, the sample is a lactate sample or a Ethanol sample.

This application is particularly important in biomedicine.  

It is particularly preferred if the first and the third high frequency pulse one x phase and the second and third Radio frequency pulse have a y phase.

According to another particularly preferred embodiment the invention becomes volumenselek in a manner known per se tive representation of the sample of a sequence of gradient magnet field pulses exposed to different coordinate directions, and at least one radio frequency pulse becomes slice selective set.

It is further preferred if in the middle of the time between the first and the second high-frequency pulse a 180 ° pulse is irradiated onto the sample.

This measure has the advantage that the evolution of the spins refocused due to a shift in chemical shift becomes.

Advantageously, after the fourth high-frequency pulse a second 180 ° high-frequency pulse is radiated onto the sample, preferably after a time interval of 1/4 year.

This measure has the advantage that in the time interval between the fourth high-frequency pulse and the signal recording Evolution of the spins is also refocused.

In a further preferred variant of the invention The time interval between the first high frequency pulse and the second high-frequency pulse equal an odd multiple of the reciprocal of double  the coupling constant between the first group and the second group.

Finally, a particularly good effect is achieved if in a manner known per se for imaging a volume-selective measurement of a plurality of areas the sample is carried out.

Further advantages result from the description and the attached drawing. It is understood that the above mentioned and the features to be explained below not only in the specified combination, but can also be used in other combinations or alone without departing from the scope of the present invention.

Embodiments of the invention are in the drawing are shown and are described in more detail in the following description explained. Show it:

Figure 1 is a diagram for explaining a pulse sequence, as used in the method according to the invention.

Fig. 2A is a 200 MHz proton spectrum of 100 mM lithium lactate, and ethanol in 10% D ₂ O / H ₂ O when measured with a 90 ° pulse and FID recording;

Fig. 2B shows the spectrum of Figure 2A, but with editing of the lactate CH₃ using zero quantum coherence filtering.

FIG. 2C shows the spectrum of FIG. 2A, but using an ECZOTIC pulse sequence according to FIG. 1, frequencies and pulse durations having been set for editing lactate;

FIG. 3A again shows a spectrum like FIG. 2A for comparison purposes;

Fig. 3B shows the spectrum of Fig. 2A, but with editing of the ethanol CH₃ using zero quantum coherence filtering;

... Figure 3C, the spectrum of Fig 2A, but using a ECZOTIC pulse sequence according to Figure 1, with frequencies and pulse widths have been adjusted for editing ethanol;

4A is a spectrum recorded with a ECZOTIC-in pulse sequence for lactate, with phase alternation of the 90 ° pulse for forming a Multiquanten- coherence.

FIG. 4B is a spectrum similar to Fig. 4A, but for ethanol with phase of the initial 90 ° pulse.

Generally speaking, it is an object of the present invention to manipulate a spin system X _n Y _m , where X and Y are proton spins with scalar coupling. Within a high field approach, the initial spin coherences can be determined by a product operator

U ⁿ S _z V ^m-1 (for Y _m ) and
U ^n-1 I _z V ^m (for X _n )

to be discribed. U and V are unit operators, like this e.g. in the US Journal of Magnetic Resonance, 55 (1983), page 128. Assuming that m is n, for the signal Coherence of interest to noise ratio:

U ⁿ S _z V ^m-1

Fig. 1 shows a so-called ECZOTIC pulse sequence. For the following consideration, a spin system with n = 1 is assumed, as occurs, for example, in lactate. A first 90 ° pulse 10 is selectively irradiated at the resonance frequency ω _{s of} the CH₃ in order to maximize the subsequent suppression of the water. The resulting changes in spin coherence after the 1 / 2Y evolutionary period are as follows:

Here, π / 2 (ω _s , x) is a selective "soft" high-frequency pulse 10 with + x phase, the center frequency of which lies on the resonance frequency ω _s . It is sufficient if narrow-band pulses with a Gaussian shape are used.

After a time t of 1 / 4J, a 180 ° high-frequency pulse 11 is now radiated in, which is a "hard" high-frequency pulse and is used to refocus the evolution due to a shift in the chemical shift. The 180 ° high-frequency pulse 11 can also be set disc-selectively, but not selectively with regard to the chemical shift.

A second selective 90 ° high-frequency pulse 12 follows after a further time interval 1 / 4J. The second selective 90 ° high-frequency pulse 12 is in the + y phase and is selectively radiated onto the resonance frequency ω _{I of} the lactate CH. The second 90 ° high-frequency pulse 12 generates a multi-quantum coherence, which is then immediately converted into a correlated z-order, namely by a suitably set "hard" 90 ° high-frequency pulse 13 , which is immediate connects to the second "soft" 90 ° high-frequency pulse 12 . This "hard" 90 ° high-frequency pulse 13 can also be set in a disk-selective manner. The result is:

At this point, all other spins, including water, have only received the "hard" 90 ° RF pulse 13 , which produces single quantum coherence. If pulsed magnetic field gradients 15 are now exerted on the sample, this transverse coherence dephases, the correlated z-order of the CH₃ group of the lactate, I _z S _z V ^m-1 , being unaffected. It is known from US-Z "Journal of Magnetic Resonance", 80 (1988), page 168 to generate a correlated z-order, but in different ways, because there is an experiment with a correlation of the occupation transfer using the phase rotation characteristic of half-Gaussian pulses outside of the resonance was used.

The pulse sequence of FIG. 1 can now be continued from different points of view in order to convert the correlated z-order into an observable signal.

According to a first variant of the pulse program, a selective 90 ° high-frequency pulse 14 is radiated, the center frequency of which is at the resonance frequency ω _s and generates a single-quantum coherence for the CH₃ spins in a state with J-order, which is easier Is refocused by waiting a further period 1 / 2J before a signal 16 is recorded, ie:

It should be noted that the final signal of the total available spin coherence of the CH₃ group corresponds.

According to another variant of the pulse sequence according to FIG. 1, pulse 14 is irradiated at the end as a selective read pulse of frequency ω _I , which generates a single-quantum coherence for the CH spins by means of polarization transfer from the CH₃ group, but with phase and Intensity distortions occur:

Furthermore, a further 180 ° high-frequency pulse 17 can also be radiated in 1/4 time after the fourth high-frequency pulse 14 , which is also set to “hard” or disk-selective. This pulse 17 refocuses the evolution of the spins due to a shift in the chemical shift.

FIG. 2 shows a comparison between the effects of the ECZOTIC pulse sequence according to FIG. 1 and a spectrum editing with zero quantum transfer in the case of lactate. It can be predicted that the signal generated by zero quantum coherence selection is smaller by a factor of 4 than that generated by an ECZOTIC pulse train according to FIG. 1, because of the effect of the pulsed magnetic field gradients on multi-quantum coherence, as mentioned above, and the formation of a mixture of zero-quantum components with x and y phases, only one of which can be converted into one-quantum coherence at a time.

The measurement of Fig. 2 was recorded at 200 MHz proton frequency on a sample of 100 mM lithium lactate and ethanol in 10% D₂O / H₂O. Fig. 2A shows a conventional spectrum with a single signal recording using a normal 90 ° rf pulse of x-phase. Fig. 2B shows a spectrum by editing the lactate CH₃ with zero quantum coherence filtering, and Fig. 2C shows a corre sponding spectrum, which was recorded using the pulse train according to FIG. 1, the frequencies and pulse durations for editing the Lactats were measured.

From these measurements it can be seen that the signal-to-noise ratio of the measurement according to the invention according to FIG. 2C is approximately three times greater than that which can be obtained with a zero quantum editing. The excellent suppression of the water signal by a factor of at least 4000: 1 allows the signal receiver to be operated with a high amplification factor in order to obtain an optimal signal-to-noise ratio.

If the ECZOTIC pulse sequence according to FIG. 1 is now applied to a spin system with n = 2, for example to ethanol, the following consideration applies: The 90 ° phase shift of the second 90 ° high-frequency pulse 12 to form a multi-quantum coherence at ω _I and the following "hard" 90 ° high-frequency pulse 13 is not necessary because of the evolution of the S-spins of the CH₃ group coupled with the two I-spins. One finds:

In this case, too, the correlated z-order can be converted back into an observable single-quantum magnetization for the CH₃ spins with maximum signal yield, or for the CH₂ spins by means of polarization transfer, depending on the frequency of the last selective pulse. The greatest possible suppression of water in the former case can be ensured by the selectivity of the Gaussian reading pulse with the frequency ω _s .

The difference in signal to noise ratio between that ECZOTIC method and zero quantum filtering can be used can be predicted by a factor of 8. In addition to that factor 4 already mentioned above for the case n = 1, is therefore another factor of 2 in the step of training the Multi-quantum coherence lost when n = 2. Because it will both single-quantum and three-quantum components are generated and then leave during the period of effect lost from spoil gradients.

A comparison of the two methods when editing ethanol is illustrated in Fig. 3, where Fig. 3A shows a normal spectrum with a sample as in the measurement of Fig. 2A, Fig. 2B shows an edited ethanol CH₃ spectrum using Shows zero quantum coherence filtering and FIG. 3C finally illustrates the application of the ECZOTIC pulse sequence with excitation of ethanol. It can be seen that the signal-to-noise ratio of the CH₃ resonance is better by a factor of 6.

The dependency of the signal edited by means of an ECZOTIC pulse sequence on the relaxation time T₂ only arises during the two 1 / 2J evolution time intervals, while in methods with zero quantum filtering the coherence of interest is subject to a T₂ relaxation, namely during the entire pulse sequence , which consists of at least three 1 / 2J time intervals. The relatively long homospoil time of 170 milliseconds is due to the low gradient field strength of the spectrometer used in the measurements. All remaining error signals which are formed by T ₁ relaxation can be removed by setting an addition / subtraction phase cycle on the first selective 90 ° high-frequency pulse 10 in interaction with the amplifier. In general, it can be said that the shorter length of the ECZOTIC pulse sequence reduces the dependence on T ₁ and T ₂ in comparison to a zero-quantum coherence transfer sequence.

An essential feature of spectroscopy with correlated z-order is the effect of the phase rotation of the pulse 12 to form a multi-quantum coherence and the hard pulse 13 to form the z-order. If n = 1, a phase rotation of the π / 2 (y, ω _I ,) pulse inverts the sign of the multi-quantum coherence and thus provides a mechanism for an addition / subtraction cycle to eliminate remaining error signals.

Fig. 4A shows the result of such Doppelakquisitions- phase cycle for lactate. However, as can be seen from a consideration of the analysis of the product operators explained above, a phase reversal of the following 90 ° high-frequency pulse 12 with y-phase does not change the phase of the correlated z-order because the pulse 12 has the signs of both I _x and also reverses from S _x . The opposite is true when n = 2 because the reversal of the selective radio frequency pulse 12 to form the multi-quantum coherence does not reverse the sign of the multi-quantum coherence because it is applied to an Iy² component. A phase change of the hard 90 ° high-frequency pulse 15 with x-phase in interaction with the receiver is now feasible. An even more effective removal of the remaining lactate signal could still be obtained by reversing the phase of the first selective 90 ° high-frequency pulse 10 of the frequency ω _s and of the x-phase, as is illustrated in the spectrum of FIG. 4B.

The improved signal-to-noise ratio which can be achieved with the method of the present invention gives rise not only to use in editing studies of high-resolution spectra of aqueous solutions, but also in the imaging of metabolites and in volume-selective experiments in vivo . In this case, one would essentially first establish the correlated z-order for the metabolite of interest, and the homospoil field gradient 15 according to FIG. 1 could be replaced by a volume-selective sequence, the localized metabolite signal then after the last selective pulse 14 would be observed.

In the case of volume-selective presentation of lactate, however, a problem arises when a phase reversal per slice is required for an addition / subtraction cycle. In general, the correlated z-order generated by the method according to the invention is described with the expression Iz ⁿ Sz. If n = 1, as in the case of lactate, the sign of the correlated z-order cannot be reversed by a hard 180 ° high-frequency pulse or by a selective 180 ° high-frequency pulse in the presence of a slice gradient the same reasons as explained above. For these reasons, either a procedure for alternative reversal of the slices is required or the volume selection must be carried out before the spectral editing.

If n = 2, the correlated z-order is always odd and can be reversed in the usual way.

The spectra shown in FIGS . 2 to 4 were recorded on a 200 MHz spectrometer with a magnet system of 4.7T. The magnet system had a usable diameter of 13 centimeters and was provided with a deuterium field frequency lock. The selective excitation was achieved with a 90 ° high-frequency pulse of low power of 20 ms duration and Gaussian pulse shape (f (x) = exp (-ax²), a = 3.0, cut off at 1%). The duration of the hard pulse was typically 20 µs. The spectra were recorded on samples in non-rotating 10 mm NMR tubes containing 10% D₂O. The repetition rate for those spectra in which two measurements were carried out in succession was 25 s. The homospoil gradient was generated by pulsing the room temperature shim (z), which gave a maximum gradient of 2.7 mT / m.

Furthermore, volume-selective measurements can also be carried out within the scope of the present invention. For this purpose, at least one of the high-frequency pulses 11 , 13 or 17 is set in a disk-selective manner and a magnetic field gradient pulse is applied at the same time. Imaging methods can also be carried out, as is known per se. For example, one of the high-frequency pulses 11 or 13 can be set in a disk-selective manner and a phase gradient can be switched in the time interval between the fourth high-frequency pulse 14 and the signal recording. Either the signal 16 is then recorded under a reading gradient or a second phase gradient is switched in the time interval after the fourth high-frequency pulse 14 in order to obtain imaging as a function of the chemical shift imaging.

The exemplary embodiments described relate to the case of lactate, in which the coupling between the protons of the two groups CH ₃ and CH is present (homonuclear coupling). However, it goes without saying that the invention also relates to cases in which the coupling exists between different types of nucleus (heteronuclear coupling). In this case, however, the "hard" or disk-selective high-frequency pulses 11 , 13 , 17 must be replaced by a pair of simultaneous and in-phase pulses, one of which acts on one of the core types.

Claims (9)

1. A method for recording nuclear magnetic resonance spectra of samples with at least two groups (CH ₃ , CH) of nuclei of the same type of nucleus, whose chemical shift essentially coincides, the sole representation of the signal of the first group (CH ₃ ), the signal of the second group (CH) is suppressed, characterized in that

• that a pulse sequence of four high-frequency pulses, preferably 90 ° high-frequency pulses ( 10 , 12 , 13 , 14 ), is radiated onto the sample,
• - that the first high-frequency pulse ( 10 ) is a soft pulse that is selective for the cores of the first group (CH ₃ ),
• - That the second radio frequency pulse ( 12 ) is a selective, soft pulse for the cores of the second group (CH) and is set such that the magnetizations of the cores of the first group (CH ₃ ) and the second group (CH) at least partially converted to a state of multi-quantum coherence,
• - That in immediate succession after the second high-frequency pulse ( 12 ), the third high-frequency pulse ( 13 ) is irradiated as at least one hard pulse on the sample and set such that the multi-quantum coherence in a state of the correlated z-order is transferred
• - That in the time interval (τ _HS ) between the third high-frequency pulse ( 13 ) and the fourth high-frequency pulse ( 14 ) a dephasing gradient magnetic field pulse ( 15 ) is exerted on the sample ( 30 ),
• - That the fourth high-frequency pulse ( 14 ) for the nuclei of the first group (CH ₃ ) is selective, soft pulse and is set such that the state of the correlated z-order again in a state of single-quantum coherence of the nuclei first group (CH ₃ ) is transferred, and
• - That the resonance signal of the nuclei of the first group (CH ₃ ) is then detected.

2. Method for recording nuclear magnetic resonance spectra of samples with at least two groups (CH ₃ , CH) of nuclei of the same type of nucleus, the chemical shift of which essentially coincides, the sole representation of the signal of the first group (CH ₃ ) the signal of the second group (CH) is suppressed, characterized in that

• that a pulse sequence of four high-frequency pulses, preferably 90 ° high-frequency pulses ( 10 , 12 , 13 , 14 ), is radiated onto the sample,
• - That the first high-frequency pulse ( 10 ) is a selective, soft pulse for the cores of the first group (CH₃),
• - That the second high-frequency pulse ( 12 ) is a selective, soft pulse for the cores of the second group (CH) and is set in such a way
• - That the magnetizations of the nuclei of the first group (CH₃) and the second group (CH) are at least partially converted into a state of multi-quantum coherence,
• - That in immediate succession after the second high-frequency pulse ( 12 ), the third high-frequency pulse ( 13 ) is irradiated as at least one hard pulse on the sample and set such that the multi-quantum coherence in a state of the correlated z-order is transferred
• - That in the time interval (τ _HS ) between the third high-frequency pulse ( 13 ) and the fourth high-frequency pulse ( 14 ) a dephasing gradient magnetic field pulse ( 15 ) is exerted on the sample ( 30 ),
• - That the fourth high-frequency pulse ( 14 ) is a selective, soft pulse for the cores of the second group (CH) and is set such that the state of the correlated z-order via polarization transfer from the cores of the first group (CH ₃ ) is converted into a state of single quantum coherence of the nuclei of the second group (CH), and
• - That then the resonance signal of the cores of the first group (CH₃) is detected.

3. The method according to claim 1 or 2, characterized in that the sample is a lactate sample or an ethanol sample is.   4. The method according to one or more of claims 1 to 3, characterized in that the first high-frequency pulse ( 10 ) an x-phase, the second high-frequency pulse ( 12 ) a y-phase, the third high-frequency pulse ( 13 ) has a y phase and the fourth high-frequency pulse ( 14 ) in turn has an x phase. 5. The method according to one or more of claims 1 to 4, characterized in that in the middle of the time between the first and the second high-frequency pulse ( 10 , 12 ) a first 180 ° high-frequency pulse ( 11 ) is radiated onto the sample becomes. 6. The method according to one or more of claims 1 to 5, characterized in that after the fourth high-frequency pulse ( 14 ), a second 180 ° high-frequency pulse ( 17 ) is radiated onto the sample. 7. The method according to one or more of claims 1 to 6, characterized in that the sample is exposed to a sequence of gradient magnetic field pulses of different coordinate direction (x, y, z) in a known manner for volume-selective display and that at least a high-frequency pulse ( 11 , 13 , 17 ) is set disc-selectively. 8. The method according to one or more of claims 1 to 7, characterized in that the time interval between the first high-frequency pulse ( 10 ) and the second high-frequency pulse ( 12 ) is equal to an odd multiple of the reciprocal of twice the coupling constant (J ) between the first group (CH ₃ ) and the second group (CH). 9. The method according to one or more of claims 7 or 8, characterized in that in known per se Way an imaging is performed.