JPH0421492B2 - - Google Patents

Description

[Detailed Description of the Invention] (a) Industrial Application Field The present invention relates to an NMR tomography apparatus, and particularly relates to an NMR tomographic imaging apparatus, and in particular, to an NMR tomographic imaging apparatus that captures N magnetic resonance signals in N rows consisting of two sets of N' sampling points taken simultaneously. This invention relates to an apparatus for obtaining a tomographic image having N×N' pixels by Fourier transformation.

(b) Prior art In Figure 1, the body axis direction is the Z direction, and the longitudinal direction is the Y direction.
If the direction and lateral direction are the X direction, conventional NMR
In the tomographic imaging apparatus, by applying a magnetic field gradient Gx in the X direction, a high frequency wave of a constant frequency applied to the subject is frequency modulated.
A method (Kumar, Welt, Ernst's two-dimensional Fourier transform method, or There is an improved method of this method, such as the spin warp method (hereinafter referred to as the two-dimensional Fourier transform method) (Reference: ZHCHO et al., “Fourier Transform
Nuclear Magnetic Resonance Tomographic
Imaging”PROCEEDINGS OF THE IEEE,
VOL.70, No.101152-1173, OCTOBER1982).

Therefore, when calculating a tomographic image by the two-dimensional Fourier transform, in order to obtain an N×N pixel image within one tomographic plane, 2N rows of nuclear magnetic resonance signals, each row having 2N sampling points, are required. was necessary.

Therefore, since the longitudinal relaxation time of the nuclear magnetic moment is approximately 1 second, the time required to collect one row of nuclear magnetic resonance signals is approximately 1 second, which is approximately equal to the longitudinal relaxation time of the nuclear magnetic moment.

Therefore, there is a drawback that it takes a long time to acquire the required number of nuclear magnetic resonance signals. Furthermore, since the calculation time for obtaining the tomographic image becomes correspondingly longer, the time from the start of signal acquisition to the transfer of the tomographic image, that is, the total time of the signal acquisition time and the calculation time, also increases. It had the disadvantage of being stored away.

In order to eliminate such drawbacks, the present inventor obtained N rows of nuclear magnetic resonance signals each consisting of 2N' sampling points by applying a magnetic field gradient Gy determined by a predetermined relationship in the Y direction, and We have proposed a method for obtaining a cross-sectional image consisting of N×N' pixels by performing two-dimensional Fourier transform on N rows of nuclear magnetic resonance signals using a computer.
(Special Application No. 58-056880).

However, this method has the disadvantage that the S/N ratio of the tomographic image is 1/√2 of the conventional method because the amount of acquired signals is halved.

(c) Purpose This invention provides an NMR tomography apparatus that can reduce the signal acquisition time and computer calculation time by half, without reducing the S/N, compared to conventional NMR tomography apparatuses. is intended to provide.

(d) Configuration This invention provides a magnetic field gradient Gy twice the normal one in the Y direction (vertical direction), and uses two reference signals with a phase difference of 90 degrees to each other, so that each row has N' It is configured to obtain two sets of N rows of nuclear magnetic resonance signals consisting of N sampling points, and to obtain an N×N' cross-sectional image having the same S/N and resolution as the conventional device.

(E) Embodiment The above-mentioned FIG. 1 is also used to explain the present invention, and in FIG. In addition, FIG. 7B shows the cross section 2 itself taken out.

Here, the procedure for obtaining nuclear magnetic resonance signals will be explained.

That is, the cross-sectional position in the body axis (Z-axis) direction is determined by applying a magnetic field gradient Gz simultaneously with the application of a 90° RF pulse (for example, 6 Mz).

Next, in the X direction, the magnetic field gradient Gx
(However, the magnetic field gradient Gx is determined so that the resonance frequency difference is f at the positions of X=0 and X=Lx.)

Therefore, as shown in equation (1), the magnetic field gradient Gy in the Y direction is calculated for each row (L=0, 1, 2,
...Add every N-1).

Gy(L)=G ^o y(L-N/2)...(1) However, Gy is the center Y as shown in Figure 2.
A magnetic field gradient in the Y direction that becomes zero at =Ly/2, where Y=0, γG ^o y=-Π, Y=Ly, and γG ^o y=Π.

Here, γ is the gyromagnetic ratio. That is, the above means that the magnetic field gradient in the Y direction is doubled compared to the conventional one.

Under the above conditions, the generated nuclear magnetic resonance signal has a resonance frequency of X=Lx/2, and
After mixing two sets of reference signals with a 90° phase difference as shown in the block diagram of Fig. 6, which will be described later, the signal is passed through a low-pass filter.
~f/2 (0 to 5 KHz in the above example) signals, and further sampled at N points at time intervals of 1/f, which satisfies the sampling theorem.

This procedure is L=0, 1, 2,...N-1
By repeating this, two sets of N rows of signals each consisting of N sampling points are obtained.

Next, the two sets of N× obtained as above
By creating N×N complex matrix data by setting one of the N signals as the real part and the other as the imaginary part, performing a two-dimensional Fourier transform on this data, and calculating its absolute value, it is possible to obtain data from N×N pixels. Obtain a cross-sectional image.

The above principle will be explained below using mathematical formulas.

FIG. 3 shows an object (a triangle in the figure) 21 to be photographed in a cross section of the subject. Let H _p be the static magnetic field in the Z-axis direction applied to the subject. and,
If the resonant frequency at the position of X=Lx/2 (hereinafter referred to as center resonant frequency) is f', then f'=2Πω'=2ΠγH _p ...(2) Now, as shown in Fig. 3, ρ( When the above sequence is applied to an object having a _density ^of dx・∫ ^Ly ₀ dy・ρ(x,y) ×SIN[2Π{f′+f(x-Lx/2)/Lx}t +2Π(y-Ly/2)/Ly(L-N/2)] ...(3) becomes.

When this is mixed with a reference signal r1(t) having a resonant frequency of x=Lx/2, r1(t)=COS(2Πf't+θ)...(4), M1(t, L)=S(t, L)・r1(t) =1/2・∫ ^Lx ₀ dx・∫ ^Ly ₀ dy・ρ(x, y) ×SIN[2Π{2f′+f(x-Lx)/Lx}t +2Π(y-Ly /2)/Ly(L-N/2)+θ] +SIN[2Πf(x-Lx/2)/Lxt +2Π(y-Ly/2)/Ly(L-N/2)-θ]} ……( 5) Here, if we pass M1(t, L) through a low-pass filter and remove the first term in equation (5), we get the signal S′1(t, L) in equation (6).
L) is obtained. That is, S'1 (t, L) = ∫ ^Lx ₀ dx・∫ ^Ly ₀ dy・ρ(x, y) ×SIN [2Πf (x−Lx/2)/Lxt +2Π(y−Ly/2)/Ly (L-N/2)-θ}...(
6) This is the interval that satisfies the sampling theorem, Δt=
When sampling at N points at 1/f, equation (7) is established from t=k/f (k=0, 1, 2, . . . N-1). That is, S″1 (k, L) = ∫ ^Lx ₀ dx・∫ ^Ly ₀ dy・ρ(x, y) ×SIN[2Π(x−Lx/2)/Lxk ×2Π(y−Ly/2)/ Ly(L-N/2)-θ〕……(
7) Here, k=0, 1, 2,..., N-1 L=0, 1, 2,..., N-1.

Furthermore, as shown in Fig. 3, if the object is discretized by dividing x (0~Lx) and y (0~Ly) into N, then x=mLx/N, m=0, 1,... Using N-1 y=nLy/N, n=0, 1,...N-1...(8), it is expressed as in equation (9).

S1 (k, L) = _N-1 〓 ^m=0 _N-1 〓 ⁿ⁼⁰ ρ (m, n) ×SIN [2Π (m/N-1/2) k +2Π (n/N-1/2 )(L-N/2)-θ〕……(
9) However, k = 0, 1, 2, ..., N-1 L = 0, 1, 2, ..., N-1 ... (10) In general, N is a number around 128 or 256. It is normal to have some.

Now, as another reference signal r2(t), a signal having the same frequency as r1(t) but delayed in phase by 90° is used.

That is, r2 (t) = COS (2Πf't + θ - Π / 2) = SIN (2Πf't + θ) ... (11) For the nuclear magnetic resonance signal S (t, L) expressed by equation (3) , r2(t), passes through a low-pass filter, and performs N-point sampling at Δt=1/f. In the same way as the calculation above,
Obtain S2 (k, L) expressed by equation (12).

S2 (k, L) = _N-1 〓 ^m=0 _N-1 〓 ⁿ⁼⁰ ρ (m, n) × COS [2Π (m/N-1/2) k +2Π (n/N-1/2 )(L-N/2)-θ]...
(12) However, k and L are expressed by equation (10).

Next, create a complex matrix s(k,L) using equation (13).

s(k,L)=S1(k,L)×j +S2(k,L)...(13) Then, s(k,L) is expressed by equation (14).

s (k, L) = _N-1 〓 ^m=0 _N-1 〓 ⁿ⁼⁰ ρ (m, n) × Exp [{2Π (m/N-1/2) k +2Π (n/N-1/ 2) (L-N/2)-θ}j] ... (14) However, k = 0, 1, ..., N-1 L = 0, 1, ..., N-1 ... (15) The following The procedure for obtaining s(k,L) is shown in FIG.

Now, from the periodicity of the discrete Fourier transform, the object ρ
(m, n) is other than 0≦m<N, 0≦n<N,
Regarding m and n, each may have a period of N.

Here, similarly, a function ρ'(m, n) having period N with respect to m and n is defined by equation (16).

ρ′(m, n)=ρ(m+N/2, n+N/2) (16) Then, s(k, L) can be transformed as follows.

s (k, L) = _N-1 〓 ^m=0 _N-1 〓 ⁿ⁼⁰ ρ' (m, n) × Exp [{2Π(mk/N+2Πn/N × (L-N/2)-θ} j〕 = _N-1 〓 ^m=0 _N-1 〓 ⁿ⁼⁰ [{ρ′(m,n) ×Exp(−Πnj)Exp(−jθ)}× Exp{2Πmk/N+2ΠnL/N} =N ² D ^-1 _N,N {ρ′(m,n) ×Exp(−Πnj)Exp(−jθ)} ...(17) However, D _N,N {g(k,L)} is k and L represents the N-point discrete two-dimensional Fourier transform for
D ^-1 _N,N {k(k,L)} is N for m and n
Let it represent the discrete two-dimensional inverse Fourier transform of a point.

From equation (17), we obtain the following equation.

D _N,N {s(k,L)}=N ² ρ′(m,n) ×Exp(−Πnj)Exp(−jθ)} …(18) If we take the absolute value of equation (18), We get the following equation.

|D _N,N {s(k,L)}| =N ² ρ(m-N/2, n+N/2) (19) The image obtained by equation (19) is shown in FIG. m, n
has a period N, m'=m+N/2 n'=n+N/2 (20) If we transform Figure 5, we can obtain an image of the object with the density distribution shown in Figure 3.

FIG. 6 shows a circuit block diagram for realizing the above-mentioned NMR tomography apparatus and extracting two sets of nuclear magnetic resonance signals arranged in N rows and N columns and inputting them into a computer.

That is, the 90° RF pulse generating circuit 20 is driven by the reference clock generating circuit 10, generates a 90° RF pulse, and enters the transmitting/receiving switching circuit 40 via the RF amplifier 30, and the transmitting/receiving switching circuit 40 is connected thereto. Send a 90° RF pulse to the RF coil 50.

Based on the magnetic field gradient Gz, Gy(L) when L=0 in the above equation (1) is added, and Gx is further added, one line of nuclear magnetic resonance signals is obtained, and the transmission/reception switching circuit 40, preamplifier 60
The signal is inputted to two sets of mixer circuits 70 via.

On the other hand, the reference clock from the reference clock generation circuit 10 enters the reference signal circuit 80,
Here, two shift signals having mutually different phases by 90 degrees are generated. These two signals are input to the mixer circuit 70 respectively.
The output from is inputted to a low pass filter 90 as an analog signal including N sampling points,
The output of the A/D circuit 100 is input to the computer 110 as the first row of data.

By performing the same operation as above, two sets of L
=1, 2...N-1 rows of data are stored in the computer 11
Enter 0.

The computer 110 performs the above-mentioned two-dimensional Fourier transform based on N rows of data consisting of two sets of N sampling points, and calculates ρ as shown in FIG.
A desired image having a distribution density of (x, y) is displayed on a separate playback device.

Further, in the above-described embodiment, a square image such as N×N is described as the pixels to be obtained, but in actual practice, an image to be examined of a subject may be rectangular. Therefore, the number of data included in one scan is not limited to the above N, but is generally
It can be treated as N′.

(F) Effect The NMR tomography apparatus according to the present invention is an apparatus that obtains nuclear magnetic resonance signals by frequency modulating the X direction of the tomographic plane and phase modulating the Y direction, and in which a predetermined signal is obtained in the Y direction. The magnetic field gradient Gy determined by the relationship
By adding and mixing two reference signals with a phase difference of 90 degrees,
N consisting of two sets of N sampling points each
Because the nuclear magnetic resonance signals of the rows are simultaneously captured and a two-dimensional Fourier transform is performed on this signal,
Compared to conventional NMR tomography equipment, signal acquisition time can be cut in half without reducing S/N.

[Brief explanation of drawings]

In Figure 1, the body axis direction is the Z direction, the longitudinal direction is the Y direction,
FIG. 2 is an explanatory diagram showing the positional relationship between the subject and its tomographic image when the horizontal direction is the X direction, which is the present invention.
An explanatory diagram showing the Y-direction distribution of the Gy gradient in the NMR tomography device. Figure 3 is the object to be imaged in the cross section of the subject (triangle in the figure)
Figure 4 is a diagram to explain the processing procedure of the nuclear magnetic resonance signal to be collected, Figure 5 is a diagram showing the value of the above equation (18) as an absolute value, and Figure 6 is the In order to realize a NMR tomography device with
A block diagram of a circuit for extracting a set of N rows of nuclear magnetic resonance signals and inputting them into a computer is shown, respectively.

Claims (1)

[Claims] 1 By applying a magnetic field gradient Gx in the X direction of the subject, a high frequency wave of a constant frequency given to the subject is frequency modulated, and on the other hand, by applying a magnetic field gradient Gy in the Y direction, the high frequency wave is phase modulated. In an NMR tomographic imaging device that obtains a tomographic plane of a subject based on the each
Two sets of nuclear magnetic resonance signals in N rows made up of N' sampling points are acquired simultaneously, and N×N' complex matrix data with one set as the real part and the other set as the imaginary part is created, and this is processed by a computer. An NMR tomography apparatus characterized in that it obtains a cross-sectional image consisting of N×N' pixels by performing two-dimensional Fourier transformation.