JPH04158836A - Magnetic resonance diagnosing apparatus - Google Patents

Abstract

PURPOSE:To improve a localization characteristic and an S/N ratio with a higher coincidence between a measuring area of spectroscopy and a lesion part by obtaining a magnetic resonance signal of the same specified nuclide from a plurality of specified parts adjacent to one another of an object to be inspected to add up. CONSTITUTION:A magnetic resonance signal of phosphorus is obtained through a probe 13 for <31>P and a transmitting/receiving section 17 for <31>P. A four-dimensional reverse Fourier transform is performed with a reconstruction processing section 21. At an addition processing section 22, a spectrum of a voxel corresponding to a lesion part is determined by a phase processing and spectral components in areas are added up to obtain a spectrum in the entire area, which achieves a higher coincidence ratio between a measuring area and the lesion part. On the other hand, when uniformity of an electrostatic magnetic field is poor, a magnetic resonance signal of <1>H is obtained through a probe 14 and a transmitting/receiving section 18 both for <1>H to be sent to a magnetic field uniformity measuring section 24 via a data collection section 19. The spectrum of <1>H is measured just as that of the phosphorus done. A frequency shift value of the phosphorus spectrum is determined to correct by a peak frequency H of a water signal with a high S/N ratio thus obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a magnetic resonance diagnostic apparatus, and more particularly to magnetic resonance spectroscopy.

(Prior Art) In recent years, devices that take various tomographic images of living bodies using the proton magnetic resonance phenomenon have been developed and are being used in clinical settings. Furthermore, recently, magnetic resonance spectroscopy (Magnetic Resonance Spectroscopy), which takes advantage of the inherent characteristics of magnetic resonance, measures the types and concentrations of compounds such as phosphorus, hydrogen, and carbon in living organisms based on their magnetic resonance signals.
Research and development of ce5pectrO5cOpYs (hereinafter abbreviated as MR5) is underway.

The measurement targets of MR5 are generally compounds such as hydrogen, phosphorus, and carbon, and the concentration of these compounds in living organisms is several m M /
Very low, around L. In order to measure these low concentration compounds, it is necessary to improve the S/N ratio of the probe and the like and to widen the measurement area as much as possible. However, n.j.
If the fixed region is made too wide, the spectrum of the region to be measured (lesioned region) and the spectrum of other regions (normal region) will mix, making it impossible to make an accurate diagnosis. For this reason, the MR8 uses a technology to match the measurement area with the lesion as much as possible, that is, low force erosion (Loc).
a l 1zat Jon) becomes important.

Traditionally, the magnetic field focusing method has been used as a low-force activation method.
Surface coil method, DRESS method, selective saturation method, St
Although various methods have been developed, such as the imulated Echo method, the 15IS method, and the Fourier method, none of the methods has sufficient characteristics as described below.

The magnetic field focusing method improves the magnetic field uniformity only in the measurement area,
The purpose is to measure the signal only in the measurement area, but
It is difficult to accurately control the magnetic field distribution with good uniformity only in the measurement region, and the low-force erosion characteristics are poor. For this reason, it is hardly used at present.

The surface coil method uses the sensitivity characteristics of the probe to
Although this is a method for specifying the II constant region, it is not possible to improve the sensitivity of only a specific region, and the localization characteristics are poor.

The DRESS method uses gradient magnetic fields and selective excitation pulses in combination to improve the low force erosion characteristics of surface coils. Therefore, although the low force erosion properties in the gradient magnetic field direction are improved, the localization properties in other directions are still poor.

The selective saturation method, the stimulated echo method, and the 15IS method all use gradient magnetic fields in three directions and selective excitation pulses, and the specificity of the measurement region is good. However, the measurement area of these methods is a rectangular parallelepiped, and it is difficult to cover the entire lesion.

The Fourier method adds a phase depending on the position to a signal using gradient magnetic fields in three directions, and by using this method, spectra of multiple points can be obtained simultaneously. However, even in this method, the measurement area is a rectangular parallelepiped, and it is difficult to cover the entire lesion.

(Problems to be Solved by the Invention) As mentioned above, in conventional spectroscopy, it is difficult to match the measurement area with the lesion, resulting in poor low-force distortion characteristics, and the measurement area is too narrow, resulting in poor S
There were problems such as a low /N, and a problem that the measurement area was too wide and signals from areas other than the lesion were mixed in, making it impossible to make an accurate diagnosis.

The present invention provides a magnetic resonance diagnostic apparatus that has high congruence between the spectroscopy measurement area and the lesion, provides good low-force erosion characteristics, and can obtain the spectrum of the lesion with high S/N. The purpose is to

[Configuration of the Invention (Means for Solving the Problems) In order to solve the above problems, the present invention applies electromagnetic waves of a specific frequency and a gradient magnetic field to a subject placed in a static magnetic field. It is characterized by acquiring magnetic resonance signals of the same specific nuclide from a plurality of adjacent specific parts of a specimen, then adding up the magnetic resonance signals from these plurality of specific parts, and displaying the spectrum of the added magnetic resonance signals. shall be.

Furthermore, if there is magnetic field non-uniformity in the static magnetic field, a mutual frequency shift will occur between the magnetic resonance signals from the plurality of specific locations, so it is desirable to further include means for correcting this. Specifically, this correction means is, for example, a frequency difference between a plurality of specific parts of a spectral peak with the largest magnetic resonance signal intensity, a frequency difference of proton magnetic resonance signals from a plurality of specific parts, or a plurality of specific Correct the frequency deviation based on the magnetic field distribution of the area.

Furthermore, in the present invention, setting of a plurality of specific regions is performed using, for example, a proton image of the subject. Specifically, a specific part of the proton image of the subject is specified, and regions of a plurality of specific parts are obtained using a region enlarging method. Furthermore, it is desirable to further include means for changing the positions of the plurality of specific parts so as to match the designated area on the proton image.

(Effect) When magnetic resonance signals of the same specific nuclide are acquired from multiple adjacent specific regions (voxels) of the subject and summed, the congruence between the spectroscopy measurement area and the lesion area is determined based on the overall measurement area. In this case, the improvement is much greater than when using one or a few large voxels.

In addition, due to the above addition process, the intensity of the magnetic resonance signal of the same nuclide from each voxel, which can be regarded as a path-to-space vector, increases approximately in proportion to the number of added voxels, whereas the uncorrelated noise component increases from the added voxel. increases in proportion to the square root of the number.

Therefore, the S/N of the magnetic resonance signal spectrum increases approximately in proportion to the square root of the number of added voxels, and the magnetic resonance signal spectrum of the lesion is measured with a high S/N.

(Example) Hereinafter, the present invention will be explained in detail using the drawings.

FIG. 1 is a block diagram of a magnetic resonance diagnostic apparatus according to a first embodiment of the present invention.

A subject (not shown) is arranged in a static magnetic field magnet 10 made of a superconducting magnet, and a shim coil 11 and a gradient coil 12
,! A +p probe 13 and an IHH probe 14 are arranged.

'31p probe 13 and 1H probe 14 are:
These are for obtaining magnetic resonance signals of phosphorus (31p) and proton (1H), respectively. Shim coil 11 and gradient coil 12 are driven by shim coil power supply 15 and gradient coil power supply 16, respectively. ! +p probe 13 and jHH probe 14 each have 31
The P transmitting/receiving section 17 and the IH transmitting/receiving section 18 are connected. Probe 13.1 in the transmitting/receiving section 17.18
The magnetic resonance signal received via the sensor 4 is detected, sent to the data acquisition section 19, and A/D converted. The gradient coil power supply 16 and the transmitting/receiving section 17 for 31p are controlled by a sequencer 20.

The reconstruction processing unit 21, the addition processing unit 22, and the data display unit 23 are elements for reconstructing phosphorus magnetic resonance signals of a plurality of voxels and then adding and displaying the results, and are controlled by the computer 27. be done.

The magnetic field uniformity measuring section 24 is for measuring the uniformity of the static magnetic field formed by the static magnetic field magnet 10, and sends the measurement results to the computer 27.

Both image reconstruction processing units 25 for IH reconstruct images of proton magnetic resonance signals, and the RO1 position setting processing unit 2
6 is ROI (ReglonOrInterest;
This process performs processing for setting a region of interest (region of interest), that is, a region from which phosphorus magnetic resonance signals are to be obtained. These 1H
image reconstruction processing unit 25 and RO1 position setting processing unit 2
6 is controlled by a computer 27.

Next, the operation of this embodiment will be explained.

First, the sequence shown in FIG. 2 is executed, and the 31p probe 13 and ! A magnetic resonance signal of phosphorus is acquired via the +p transmitter/receiver 17. That is, an electromagnetic wave having a resonance frequency of phosphorus is applied to the subject by the probe 13 as a high-frequency magnetic field RF, and then a gradient magnetic field Gx in three orthogonal directions is applied from the gradient coil 12 driven by the gradient coil power supply 16.
, GY, and Gz are applied, and the distance and direction from the center of the gradient coil 12 are encoded as phase information of the magnetic resonance signal. The obtained magnetic resonance signal is the same probe 1.
3 and the transmitting/receiving section 17, A/D conversion is performed at the data collecting section 19, and then four-dimensional inverse Fourier transform is performed at the reconstruction processing section 21.

As a result, the reconstruction processing unit 21 obtains a magnetic resonance spectrum of phosphorus in a region corresponding to a plurality of voxels shown in FIG. In a large voxel of about 5 cm square as shown by the broken line in FIG. 3, a spectrum with good S/N as shown in FIG. 4(a) can be obtained. In such large voxels, low force erosion becomes worse in the case of a lesion of about 2 to 3 cm in size, as shown by diagonal lines in Figure 3.
Since signals from normal brain tissue other than the lesion are mixed in, it is difficult to examine the state of the lesion based on changes in the spectrum. In order to solve this problem, the voxel size must be reduced as shown by the solid line in FIG. However, in such small voxels, the concentration of phosphorus compounds in the living body is as low as several mM, so the S
/N resulting in a bad spectrum.

Therefore, in the present invention, the addition processing unit 22 calculates the spectrum of the voxel corresponding to the lesion by phase processing,
Add according to the formula below.

p (D −Σp (xi, yi, zi, f)
(1) ρ Nispin density function f: Magnetic resonance frequency xi, yi, zi: Center position of voxel The phosphorus magnetic resonance signal of each voxel has almost the same spectrum, although there are variations depending on the living tissue, so such By addition, the signal intensity increases approximately in proportion to the number of added voxels. On the other hand, since the noise of each voxel is uncorrelated, the noise increases in proportion to the square root of the number of added voxels. Therefore, the S/N of the phosphorus spectrum after addition increases approximately in proportion to the square root of the number of added voxels.

This point will be explained in detail with reference to FIGS. 10 to 12.

FIGS. 10(a) and 10(b) conceptually show the relationship between the lesion and the measurement area in the conventional example and in the case of the present invention. 10th
As shown in Figure (a), in the conventional method, the diagonally shaded measurement area is a rectangular parallelepiped (the figure is shown as a plane), so the matching rate between the measurement area and the lesion area surrounded by a circle is low. Here, the match rate can be defined as shown in the following formula (see FIG. 11).

-Volume of B/(Volume of A+B+C) On the other hand, in the present invention, as shown in FIG. 10(b), spectra of a plurality of regions indicated by diagonal lines arranged to cover a wide range of the lesion are obtained. By adding the spectral components of each of these regions, the spectrum of the entire region can be obtained. FIG. 10(b) shows the case when eight regions are added, and the matching rate is higher in the apparatus according to the present invention than in the conventional method. If the number of measurement regions is increased, the matching rate will further increase.

Now, regarding the improvement of S/N by addition, let us consider the case where the spectra of two regions 1.11 are added as shown in FIG. 12. When the signal component is 81 and the noise component is N, the S/N (SNR+) of only measurement area I is (Sa + 5b) 8NR゛-Jπbowπbow (3) Also, when adding the spectra of measurement regions I and H S of
/N (= 5NRI+2) is SNR+-, where S''' is the signal per unit volume of the lesion, N1 is the noise, and Va, Vb, and Vc are the volumes of each area A, B, and C in Fig. 12. S”/N” * (Va+Vb)SNR
+. 2mygo'1""2Vb+Vc) (5)(Va+
4 Vb+ We) If the voxel size is V, then 5NRI -581N1* JT (
6) From the above, the smaller vb is, that is, the less the measurement areas overlap, the better the S/N is. For example, when the measurement areas do not overlap at all, the S/N increases by a factor of f7, and when they all overlap, the S/N does not change at all.

On the other hand, when the uniformity of the static magnetic field formed by the static magnetic field magnet 10 is poor, for example, if the 5th (a) is the phosphorus spectrum of a certain voxel, the phosphorus spectrum of other voxels is
As shown in Figure (b), the magnetic resonance frequency differs for each voxel.

Therefore, in this embodiment, the IH magnetic resonance signal is acquired via the IH probe 14 and the transmitting/receiving section 18, and sent to the magnetic field uniformity measurement section 24 via the data acquisition section 19. Measure the spectrum. From the peak frequency of the main signal with a good S/N obtained in this way, the frequency shift value of the phosphorus spectrum is determined and corrected. That is, the peaks of the phosphorus spectrum and water spectrum are determined for each voxel, and the peak frequency r is determined. Frequency correction is performed in advance using the following equation (8) using =rt.

p (xi, yl, zi, r) = p(xl, yl, z
i, f'÷Δr)ρ': Spin density function after frequency correction Δf: Corrected frequency (ΔI-(ro-'1)γp/yu)1P: Magnetic resonance ratio of phosphorus (magnetic resonance frequency divided by magnetic field strength) γ, : Proton magnetic resonance ratio Specifically, the frequency of the electromagnetic wave transmitted from the probe 13 and the frequency r of the reference wave for synchronous detection of the magnetic resonance signal may be shifted by Δ1.

Note that instead of performing frequency correction based on the proton resonance frequency in this way, if the S/N of the phosphorus spectrum is good, correction may be performed using the difference in frequency between the phosphorus spectrum peaks.

Alternatively, the magnetic field distribution may be measured by the Fourier method, the Dixon method, or the like, and the frequency may be corrected using the measured value as shown in the following equation (9).

p (xi, yi, zl, f') - p (xi, yi, z
l, f + Δ') Δr: correction frequency (Δr-yp ・ΔB) ΔB = relative magnetic field (9) In addition, in this example, as a method of specifying the voxels to be added, We used a method in which the positions of voxels determined by calculation are displayed and the operator instructs the positions one by one, but it is also possible to specify the vertices of a polygon and set the voxels included in that polygon as the addition area. Alternatively, the center and radius may be specified and the voxels included in that area may be added to the area.

Further, in this embodiment, phosphorus was used as the target nuclide for spectroscopy, but other nuclides such as protons and carbon may be used.

Furthermore, in this example, a sequence of Fourier transform method was used in which spectra of multiple points can be obtained in a series of sequences, but data of one point can be obtained in a series of sequences.
Similar processing can be performed using the 518 method or the 5tfIlated Echo method, which allows one point of data to be obtained in one sequence.

Next, a second example will be described. In the second embodiment, voxels are specified using a region expansion method. FIG. 7 is a schematic diagram of a tomographic image of the subject's head, and there is a tumor-like lesion 72 within the acid 71. When a histogram of the signal intensities of the magnetic resonance signals is drawn along the straight line A-8, there is a clear difference in signal intensities between the signal 81 of the brain region and the signal 82 of the lesion, as shown in FIG. In Fig. 8, the signal 8 of the lesion area is
2 is displayed as a high signal area, but depending on the disease it may become a low signal area.

Here, the operator indicates a part of the lesion with a marker,
By enlarging the region using the region enlarging method, only the lesion can be extracted. For example, when an operator indicates a certain voxel in FIG. 3 with a marker via the computer 27 and RO1 position setting processing section 26 in FIG.
The O1 position setting processing unit 26 compares the signal intensities of adjacent voxels with that of the voxel, and if the difference in signal intensities is smaller than a predetermined threshold, the voxel is identified as a lesion, and if it is greater than the threshold, the voxel is identified as a lesion. Identify as non-lesional area. Next, the signal intensities of other adjacent voxels are compared and similar processing is performed to identify whether the voxel is a lesion or not.

In this way, the signal intensities of voxels identified as a lesion and adjacent voxels are sequentially compared, and the area is enlarged. When there are no more unprocessed voxels, the RO1 position setting processing unit 26 ends the process.

As a result of such processing, in the case of FIG. 3, the shaded area in the figure is identified as a lesion. In Figure 3, 1 is used for simplicity.
Although a three-dimensional case has been described, in reality, a lesion is identified using a two-dimensional proton image and a three-dimensional proton image as shown in FIG. When the voxels included in the obtained lesion are determined by the computer 27 and the magnetic resonance spectra of the voxels are added in the same manner as in Example 1, a spectrum with good low-force distortion and S/N can be obtained.

Next, a third embodiment of the present invention will be described. When the lesion is small and the area that can be covered by voxels is narrow as shown in FIG. 9, the above two embodiments cannot expect much improvement in S/N. Therefore, by processing the magnetic resonance signal before reconstruction as shown in the formula (lO) below, the voxel is (xo, Yo
, Zo).

S (kx, ky, kz, t) −S (kx, ky
, kz, t) * exp(-j(kx x o
+ky yo +kz zo ))S
(kx, ky, kz, L): Magnetic resonance signal S after signal processing (kx, ky, kz, L): Magnetic resonance signal before signal processing kx, ky, kz: Amount of phase encoding by gradient magnetic field t: Sampling Time xo + Yo + zo: Voxel movement amount Addition in real space corresponds to addition in k space (magnetic resonance signal space), so the signal at the voxel position of ■■■■■■ in Figure 9 Addition can be performed using the following formula.

S (kx, ky, kz, t) −S (kx, ky
, kz, t)*Σexp(-j(kx x + +ky
y + +kz z I))X+, y++ z
l: Amount of movement of each voxel with respect to voxel 1 By reconstructing this signal sequence, the spectrum corresponding to voxel 1 becomes the sum of the spectra of voxel ■■■■■■.

In this example, addition was performed on k space, but
Even if the addition is performed in real space, there is no problem in principle as long as the processing takes time because the reconstruction is performed for the number of voxels to be added.

[Effects of the Invention] According to the present invention, both localization characteristics and S/N can be improved in magnetic resonance spectroscopy.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a magnetic resonance diagnostic apparatus according to an embodiment of the present invention, FIG. 2 is a diagram showing a sequence for acquiring a phosphorus magnetic resonance signal in the same embodiment, and FIG. 3 is a diagram showing a lesion area and voxels. Fig. 4 is an explanatory diagram showing the positional relationship of , Fig. 4 is an explanatory diagram of the magnetic resonance spectrum, Fig. 5 is an explanatory diagram of the frequency shift of the phosphorus spectrum due to magnetic field inhomogeneity, and Fig. 6 is an explanatory diagram to explain the frequency shift of the 1H spectrum. , FIG. 7 and FIG. 8 are diagrams for explaining the principle of the area expansion method, FIG. 9 is an explanatory diagram of area movement, and FIGS. 10, 11 and 12 are diagrams for explaining the principle of the present invention. FIG. ]O... Static magnetic field magnet, 1]... Shim coil, 12.
・Gradient coil, 13...probe for 31p, ]4・
・. 1H probe, 17...31p transmitter/receiver, 18.
... IH transmitter/receiver section 19... data collection section, 20
... Sequencer, 21 ... Reconfiguration processing unit, 224.
...Addition processing section, 23...Data display section, 24...
Magnetic field uniformity measurement unit, 25... 1H image reconstruction processing unit, 26... RO1 position setting processing unit, 27... Computer. Applicant's agent Takehiko Suzue Signal strength Figure 3 Signal strength Figure 4 Signal strength Figure 5 Signal strength f. Signal strength Fig. 6 Fig. 8 Fig. 9 (a) Conventional example (b) Present invention Fig. 10

Claims (2)

[Claims] (1) means for acquiring magnetic resonance signals of the same specific nuclide from a plurality of adjacent specific parts of the subject by applying electromagnetic waves of a specific frequency and a gradient magnetic field to the subject placed in a static magnetic field; A magnetic resonance diagnostic apparatus comprising means for adding the magnetic resonance signals from the plurality of specific sites obtained by this means, and means for displaying the spectrum of the magnetic resonance signals added by this means. . (2) means for acquiring magnetic resonance signals of the same specific nuclide from a plurality of adjacent specific parts of the subject by applying electromagnetic waves of a specific frequency and a gradient magnetic field to the subject placed in a static magnetic field; means for adding the magnetic resonance signals from the plurality of specific regions obtained by this means; means for displaying the spectrum of the magnetic resonance signals added by this means; A magnetic resonance diagnostic apparatus comprising: means for correcting frequency deviation between magnetic resonance signals from the magnetic resonance signals.