FI65859C - FOERFARANDE FOER UTREDNING AV AEMNETS KEMISKA EGENSKAPER - Google Patents

Description

65859

METHOD FOR CHECKING THE CHEMICAL PROPERTIES OF A SUBSTANCE -FÖRFARANDE FÖR UTREDNING AV ÄMNETS KEMISKA EGENSKAPER

The invention relates to a method according to claim 1 for determining the chemical properties of an object to be examined, such as, for example, the human body or a tree trunk, taking advantage of the nuclear magnetic resonance phenomenon.

5 The method is used to obtain the local distribution of the so-called chemical shift spectrum of an object.

The nuclear magnetic resonance phenomenon (hereinafter YMR) has been known since the 1940s. The first experiments were carried out by Bloch and 10 Purcell in 1946. Since then, the phenomenon has been applied in the fields of physics, chemistry and medicine.

YMR is based on the fact that the nuclei of some elements have a magnetic moment. Such are e.g. 1h, ^ F and 31p, 15 with a nuclear spin quantum number I 1/2. The magnetic moment jm of the nucleus is proportional to the spin quantum number I of the nucleus: (1) μ -t »i 20 where y · is the nucleus-dependent gyromagnetic ratio Ä = h / 2" JT; h is the Planck constant

The behavior of the nucleus in an external magnetic field can be observed by either quantum mechanics or classical mechanics. The latter is a more illustrative approach. It can be assumed that the cores are small rod magnets for which the rotation of the cores around their own axis, i.e. the spin, generates not only a magnetic moment but also a torque moment.

30

If a sample with a considerable number of, for example, 31p atoms is placed in the external magnetic field B0, most of the magnetic moments in the nuclei of said atoms are parallel to the external magnetic field and the resultant forms a so-called net magnetization parallel to the fundamental field. Said net magnetization can be deviated from the direction of the basic magnetic field by introducing electromagnetic energy into the sample at a frequency which satisfies the resonance condition; 5 (2) WQ - f Ho - 2JT / W0 * so-called Larmor frequency

The deviated magnetization prerores around the direction 10 of the fundamental field at a frequency corresponding to said Larmor frequency.

This precessing magnetization can be detected by placing an induction coil outside the target, in which a signal voltage with a Larmor frequency and proportional to the precessing net magnetic foundation is induced.

15

In order to detect the precession of nuclear magnetization by means of an induction coil, the precession of the magnetic foundation of the nuclei must take place in a phase-coherent manner, which is the situation immediately after the end of the excitation pulse. But due to the interactions between the cores of the sample 20, the cores detect slightly different magnetic fields, so their precession frequencies differ and the precession coherence deteriorates. Deterioration of coherence results in a decrease in the inducible signal and this exponential process is characterized by a relaxation time T2 (so-called spin-spin relaxation time). The deviated magnetization gradually returns to the direction of the external magnetic field Bc, i.e. the nuclear system releases into its environment the energy received during the excitation pulse. This process is also exponential in nature and is characterized by a relaxation time Τχ (so-called spin-lattice relaxation time).

The magnetic fields created by the molecule and the environment also generate several different resonant frequencies, i.e. the inducible signal has a spectrum that depends not only on the basic magnetic field B0, but also on the chemical properties of the sample such as state, molecular structure and other 65859 constituents. The effect of a sample on its own resonance spectrum is called a chemical shift, denoted by <£, which is measured as the frequency deviation from a known reference frequency in a given chemical form.

S The reference frequency is obtained by measuring the resonance signal of a known substance in the field B0 used. * The frequency deviations mentioned are parts per million of the fundamental frequency and thus S is often expressed in parts per million. In YMR spectroscopy, the sample is placed in a homogeneous magnetic field and the nuclei to be examined are excited by a high-power and short-lived radio frequency pulse. Immediately after tuning, the resonant signal induced in the signal coil is amplified, detected, analog-to-digital converted and stored with its data in memory. The computer performs a Fourier transform on the stored 15 signal, resulting in a signal spectrum. From the intensity of the components of the spectrum thus obtained and the different reference frequencies, the structure of the molecules of the substance can be deduced or the constituents in the sample can be identified by comparing the spectrum obtained with the tabulated 20 spectra. The method described is called pulse YMR spectroscopy.

In 1973, published by prof. Lauterbur was the first to apply the YMR phenomenon to the description, i.e., to map the distributions of the concentration and relaxation times of the 25 atoms under study (Nature Vol. 242, March 16, 1973, pp. 190, 191).

In the following, a method for mapping the distributions of YMR parameters is commonly referred to as the nuclear spin mapping method.

30

Numerous nuclear spin imaging methods have been developed. These have been described e.g. in the following patents: Mansfield: US 4,165,479; Garroway et al: US 4,021,726; Moore et al: 4,015,196; Hutchison et al: WO 81/02789.

These methods, as well as other nuclear spin imaging methods, are characterized by the setting of a magnetic field gradient over the imaging range of 35 65859 during signal registration. The magnetic field gradient causes the signal to be registered to also contain position information when frequency coded. On the other hand, the magnetic field gradient makes it difficult to register the spectrum of the chemical shift.

One way to collect the local distribution of the chemical shift spectrum is set forth in P. Bendel et al., Journal of Magnetic Resonance, Vol. 38, 343 ... 356, 1980.

The described method is based on the mathematical processing of signals registered during magnetic field gradients so that the spectral information can be found out.

The mapping of the YMR characteristics is based on the fact that the sample 15 is repeatedly tuned with a radio frequency pulse, after which the YMR signal is registered with the magnetic field gradient switched on. The direction of each gradient after tuning differs from the previous one and thus projections from different directions are obtained from the sample. Of the projections obtained, recon-20 is constructed, for example, by Brooks et al. (Radiology, Vol. 117, 561, 1975). The spectral information of the chemical shift can again be mathematically reconstructed based on the fact that the frequency deviation caused by the chemical shift does not depend on the direction of the external field gradient. This method requires considerably complex data processing and requires the use of projections to form an image. On the other hand, the method is not easily applicable to other known imaging methods described in addition to the above-mentioned publications by 30 mm. in Ernst: U.S. 4,070,611 and in Edelstein et al., Physics in Medicine and Biology, July 1980,

Nr. 4, ss. 751 ... 756.

Other ways to determine the local distribution of a chemical shift 35 is to selectively tune only those nuclei of the atoms of the component under study that have a particular chemical shift, and then record a 65859 inducible nuclear magnetic resonance signal. However, this method has significant drawbacks. First, in order to elucidate the spectrum of chemical shift, the imaging process has to be repeated numerous times. Second, 5 three-dimensional objects cannot be mapped because a magnetic field gradient cannot be used in the tuning phase to limit the imaging area. Of course, methods suitable for three-dimensional imaging can be used, such as those described, for example, in the aforementioned publication Ernst: US 10 4,070,611.

Further, Abe et al., U.S. Pat. No. 3,932,805? Abe et al: US 4,240,439; Damadian: US 3,789,832; Damadian: DE (OS) 2,946,847 and Sepponen: FI 58868 are known to be differentiated by an external magnetic field placed over 15 objects so that the resonance condition is met only in a limited area that can be transmitted electrically or mechanically within the object under study. The disadvantage of the methods is that if it is desired to map the local distribution of the nuclear magnetic spectrum 20 of the element under study, the target area has to be examined point by point. In this case, the examination takes a very long time and the movements of the subject, such as breathing, peristaltic movement of the intestine, etc., cause inaccuracy in the mapping. Devices based on this method are currently manufactured by e.g. Oxford 25 Instruments (England), whose "Topical Magnetic Resonance" apparatus equipped with a superconducting magnet is capable of generating a 1β spectrum from a well-defined area of a target.

It is further known to locate the region from which the nuclear magnetic resonance spectrum is measured using a suitable signal coil. In this case, the geometry of the signal coil can be used to delimit the area from which the signal is received. By using the same signal coil also as a tuning coil and changing the length and / or amplitude of the tuning pulse, the research area of the signal coil can still be mapped in a direction perpendicular to the plane of the signal coil. The weakness of the method is that when it is desired to study areas far from the signal coil, the positional accuracy deteriorates rapidly as the distance increases. The method has been used, for example, to map the 31p spectrum of the cortex, especially by means of signal coils from the surface of the skull.

5

A weakness of the above-mentioned known methods in mapping the local distribution of the chemical shift spectrum is either slow and complex data collection or complex data processing and commitment to only 10 specific imaging methods. One solution to these problems is known from GB 2,057,142 to change the direction of the magnetic field gradient repeatedly several times after the excitation pulse and to register the generated echo signals after each change of direction. This will significantly speed up data collection. However, this known method is based on the assumption that a sufficiently homogeneous basic magnetic field can be obtained, which is often not possible or at least practical due to the complexity of the equipment and the high manufacturing costs.

The object of the invention is to provide a new, simple method for determining the local distribution of said chemical shift spectrum of an object, which can be applied in several different imaging methods and which does not increase the time spent on the imaging event itself.

It is also an object of the invention to provide a method by which distortions due to inhomogeneities of the basic magnetic field can be effectively eliminated from the mapping of the chemical shift spectrum. The objects of the invention are achieved as set out in claim 1 and the subclaims. The method according to the invention enables the mapping of both two-dimensional and three-dimensional chemical shift spectra and the additional mathematical functions required by it are easily programmable on a computer and do not require a large data processing capacity. It is also an advantage of the invention that the collection of spectral information does not lead to an increase in the recording time 7 65859. A further essential advantage of the invention is the lower requirements it imposes on the homogeneity of the basic magnetic field and thus also on the equipment used with which the method is applied. This is a consequence of the fact that the inhomogeneities of the field according to the invention are first mapped using a homogeneous object, whereby the information collected from the actual object to be examined is corrected by means of the aforementioned information 10 from the inhomogeneity mapping.

The principles of the invention will now be described and the invention will be described in more detail with reference to the accompanying drawing, in which Figure 1 shows a sequence of operations of the method according to the invention.

20

Consider an object whose density distribution of the nucleus under study is Ά (X, Y, Z, δ), where X, Y, Z are the position coordinates according to the orthogonal coordinate system and 6 the magnitude of the chemical shift under consideration. For simplicity, consider a one-dimensional density distribution A (X, δ).

Referring to Fig. 1, the cores of the target are tuned with a 90 ° excitation pulse, and in step 0, when t * 0, an X-direction gradient is set over the target. The inducing signal of the tuned nuclei is (if the observation time is small compared to the natural T2 ~ value of the nuclei).

(3) S (t) - J / A (x, 6Ie1 (Wo + rgx + 6Lt dxdö 35 δχ where S (t) is the inducible signal A (x, δ) density distribution W0 Larmor frequency γ gyromagnetic ratio g field gradient δ considered spectrum frequency t is time 8 65859

When considering a signal in a rotating coordinate system with angular frequency WQ

(4) s (t). //Afx.Sje1 <Trax + SIta * a6 5 4x

The following review is performed in a rotating coordinate system.

If the direction of the gradient is reversed at time X (step 1 in turn 1), where τ is chosen such that S (t) »0 for all t> t, but x <<the natural relaxation time T2 # of the material under study is caused by the so-called spin echo? denote S] _ (t).

13 (5) s <t) - //Aix.aie11'-™* + 4, t (T9X +4) dtIdxd6 1 6x 0

If the direction of the gradient is reversed at time t * 3t (step 2 in Figure 1), a new spin echo is generated: 20 (6) S, (t) - J / A (x, 6) eltiT3x +4) t + | (T9X + e ) dt + 3x 4 *; (-rgx 46) at dJtd6

If the inversion of the gradient is performed repeatedly at 2 τ intervals, a spin echo sequence Sn (t) is obtained, which is 25 generally of the form (7) S (t) - J / A (x, 6) no [<s9 "<n> T9 * + «> * + / (Rgx + 6) · about 6x 2x dt + -jtl + sgn (n)] / (-ygx) dt + 2'Γ (η-1) 6 3 ^^ 30 where Sgn (n ) * +1 when n is even, and -1 when n is odd

Let us first consider only odd echoes to give 35 (8) S (t) - IJiWIe1 [<_r9X *, dt + n 6x - 2X (n-löl ^ g 9 65859

Perform a variable change: k = ~ Ygx +5, dx = - ^ dk.

yg

Obtain (9) S (ti «- // A * (k, 6Ieikte" ik: cei2tn6dxd6 n Ύ9 & k u

D

Perform a Fourier transform (10) F on the signal Sn. [S (tl] * τ. N igg u 10 Perform a variable change: f = 2T6, dfi = {11) .Pt [Sn (t)] - - -5 ^ / N (k, fleikxeifndf.

As can be seen, the Fourier transforms taken from the odd consecutive spin echoes differ by eif.

From the obtained Fourier transform sequence, form a Fourier series with respect to the sequence number n, i.e. 20 (12) VFttsn (t>] - * 5Tfg Ai’k’f! EiICt - Bu (k, fl

For even echoes, the same is obtained, but by making a variable change: k '= ygx + 6 (= Ygx + f / 2x) 25 (13) s_ (t) * ~ r // A' (k1, 6) eik, te "ilcTei2xn6dxd6 Ύ 6k 'g (14) Ft [sn (t) 3 * ^ And' (k ', 6) eik'X ei2xn6 and further δ * 30 (15) Fn (Ptl8n (t) l) - A'Ck', f) e1 *, i- Bg (k ', f)

By making the investments we get because A ^ (- Ygx + f = -, f) - A (x, f / 2x) A '(ygx + l =., F) = A (x, f / 2x)

35 (16) 9 2X

A (x, 6) = rYg {sg (ygx + 6,6-2De'i (Tgx *, k -Bu (- Ygx + δ, 6 · 2x) e “i (^ gx + 6 *} ίο 65859

The above analysis can be extended to multidimensional distributions Ά (X, Y, Z, δ) * The analysis can also be performed at frequencies where phase coding is performed in some direction of the coordinate system to form an image.

For such imaging methods, mention may be made of Ernst: US 4,070,611 and Edelstein et al., Physics in Medicine and Biology, July 1980, no. 4, ss. 751 ... 756. It should be further noted that although the necessary analyzes have been performed above based on the so-called Fourier analysis, it may be advantageous to use other known frequency analysis methods, especially to generate Fn (cf. Equation 12). (<10), in which case it is advantageous to use, for example, the so-called Auto-15 regressive-Burg algorithm. These other methods are summarized in Kay et al., Proceedings of IEEE Vol. 69, no. 11, 1981.

By way of example, the following describes the determination of the spectral-20 distribution in a certain part of a three-dimensional object by means of the method according to the invention with reference to Figures 2 and 3.

The three-dimensional object to be examined is placed in a homogeneous 25 magnetic field B0, the intensity of which is typically 0.04 ... 0.3 Tesla in human body imaging. The level to be examined is selected from the object by switching a field gradient over the object, for example in the direction Z according to Figures 2, i.e. a gradient g2. The strength of the field gradient is typically 5 mT / m. The selected level cores of the target are then tuned by transmitting a radio frequency tuning pulse via the transmitter / receiver coil 1, which is modulated so that its frequency band is limited. The tuning pulse is generated by a transmitter 10 under the control of the central processor 7, which 33 modulates the frequency of the basic oscillator 5 and amplifies the signal to the desired power levels. In this case, only the cores located in slice 2 are tuned.

11 65859

The slice 2 is perpendicular to the Z direction and its thickness is determined by the bandwidth and gradient gz of the tuning radio frequency pulse.

5 The most commonly used modulation formats follow a sine or Gaussian curve. The length of the radio frequency pulse is typically 10 ms and the power used is 100 W (peak).

Immediately after the excitation pulse, the direction of said 10 Z-direction gradients is reversed to correct for the non-phase caused by the Z-gradient during slice selection. At the same time, a gradient gXy orthogonal to the direction of the Z gradient, i.e. changing in the XY plane, is connected, the direction of which forms an angle fS (K) with respect to the X direction. After the time τ, the direction of the gradient gXy is reversed and the resulting spin echo Si ^ K ^ (x, y) is registered by means of a low-noise amplifier 3 connected to the transmitter / receiver coil 1. The amplified signal is detected in a quadrature detector 4, for which a reference frequency is obtained from the basic oscillator 20. The detected signal is filtered in a filter 12 and then digitalized in an analog-to-digital converter 6. after repeatedly every 2 τ time-25 and the resulting spin echo sequence Sn ^ n (x, y) is recorded (cf. Fig. 1). The sequence described above is then repeated, but the direction of the gradient gXy is changed so that it forms a different angle * 5 (K) with respect to the X direction (K is the order of the repeat). The change of direction occurs at 30 equidistant angles, where | $ S (K-1) -j6 (K) j = 180 ° / p, where p is the number of desired projections.

For the obtained signal sequences, Sr ^ lx, y) the so-called

Past Fourier transform (FFT) on processor 7 to obtain a -35 projection queue. For said project sequences, a Fast Fourier transform is performed in the direction n according to formula (12). In this case, a projection set J ^ (K) (Xry) f is obtained from the object, which can be used to reconstruct the density distribution images from the target area 12 658 59 for each transition value. The display unit 8 of the apparatus can now be shown, for example, the distribution of protons in the target with a chemical shift of 4 ppm (parts 5 per million).

Figure 3 illustrates the formation of transition distribution images by the method described above. Target region 2 is assumed to contain two nuclear depletions, one with a chemical shift of -2 ppm and the other with +2 ppm. As described above, a transition projection P5 ° (x, y) and correspondingly other projections necessary for imaging are formed P6n (x »y) · 15 The obtained projections are used to project an image of the internal distribution of the core with a certain chemical shift. A suitable method for performing back projection is described, for example, in Brooke et al., Radiology 117, 561, 1975.

20

In the above functional description, it has been assumed that the basic magnetic field used is extremely homogeneous. Practically achievable homogeneity in the imaging range is worse than 1: 106, usually 25 1: 1 () 4. Such field inhomogeneity causes similar phase shifts to successive spin echoes as the characteristic chemical shift of the test substance and thus makes it impossible to map the chemical shift.

30

Said errors caused by inhomogeneity can be corrected according to the invention by performing a mapping of an object that is completely homogeneous and using the obtained image information distorted by the field homogeneity to correct the image of the actual object. Such a correction must be performed if the inhomogeneity of the field is greater than the desired resolution in the chemical shift spectrum.

Claims (6)

13 6 5 8 5 9 The invention is in no way limited to the embodiment shown, but several variations of the invention are conceivable within the scope of the appended claims * 5 A method for determining the local distribution of the nuclear magnetic resonance spectrum of the nucleus of a test atom using YMR nuclear spin imaging methods, in which an object, e.g., a patient, is placed in a continuous, as homogeneous magnetic field as possible and subjected to: a) applying an electromagnetic excitation pulse b) at least one magnetic field gradient is placed over the target area, c) the direction of the magnetic field gradient is changed repeatedly several times, 20 d) the phases and amplitudes of the echo signals generated after each change of direction are recorded, e) a representative spectral analysis for the signal sequence in the direction of the echo sequence number 25, characterized in that said sequence of operations (ae) is first performed in such a way that the inhomogeneity of the basic magnetic field is exploited by exploiting a homogeneous object The frequency shifts due to this are mapped, so that the next 30 sequence (a-e) is performed on the subject, and then the information of the signal sequences registered from the subject in step c is corrected for the spectral shift due to magnetic field inhomogeneity from the signals recorded from the homogeneous 35. 14 05859 A method according to claim 1, characterized in that when performing said procedure sequence on the subject under test in step b a second magnetic field gradient perpendicular to said magnetic field gradient is set over the target area, the value of which is set to zero when performing steps c to e, and that - e is repeated using different amplitude values for one of the first two gradients in the magnetic field). A method according to claim 2, characterized in that during step a a third magnetic field gradient perpendicular to said two magnetic field gradients is set over the object, and that during steps c to e the value of said third magnetic field gradient is set to zero. 20 PATENTKRAV 1. The method for utilizing the local field of the nuclear magnetic resonance spectra of the atomic core under the examination of the NMR-core-response method-25 method, which is feasible, exempts patients, and does not require a magnetic field to be homogeneous. (a) the electromagnetic stimulation pulses are applied to the NMR system, 30 g & nger, d) the number and amplitude of the ecodesign of the ecodesign register.