JPH0435649A - Magnetic resonance imaging (mri) device - Google Patents

Abstract

PURPOSE:To remove DC component noise effectively by providing a means to subtract the DC component from a data after Fourier-transformed at least one-dimensionally in a multi-dimensional Fourier transformation means. CONSTITUTION:In an amplitude modulation circuit 13, a carrier wave (high-frequency wave) signal transmitted from a carrier generating circuit 16 is amplitude-modulated in the above amplitude modulated wave form, the high-frequency signal after modulation is delivered to a transmission coil 15 through a power amplifier 14, and an excitement to a body to be checked such as a human body is carried out. An NMR signal (echo signal) generated by the body to be checked such as a human body is received by a receiving coil 17, detected by a detector circuit 18, and after that, passed through a low pass filter 19, and sampled by an A/D converter 20 to be converted into a digital signal. After the carrier wave transmitted from the carrier generating circuit 16 is placed in its restored condition, a normal high-frequency signal excitement, and a pulse sequence of NMR receiving are repeated while changing a phase encoding amount, and raw data S (t,phs) of the MXM matrix are collected. And the roughly detected DC component is subtracted from the above raw data, and after that, zero- padding by M/2 point is carried out at both end in the frequency direction, so as to form 2M pieces of data, and then one-dimensional Fourier-transformation is carried out.

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to an improvement in an MRI apparatus that performs imaging using NMR (nuclear magnetic resonance).

[Conventional technology]

In an MRI apparatus, it is inevitable that the direct current component of the receiving system adversely affects reconstructed images as noise, so it is necessary to remove or correct this. At the raw data stage, which is the detected NMR signal itself,
It is possible to remove the DC component at each sampling point, but even if the remaining DC component error is small, due to the principle of Fourier transform, it will be multiplied by a matrix after transformation, so it will be corrected at the raw data stage. Therefore, conventionally, for each pixel after image reconstruction, pixels that are recognized to be an appearance of noise are interpolated from surrounding pixels, and this noise is We are trying to eliminate the negative effects of In addition, by controlling the phase of the excitation high-frequency pulse, the original DC noise is moved to the highest frequency region of the image (the farthest edge in the image phase encoding direction), and the main noise component is pushed to the edge of the field of view of the reconstructed image. Efforts are also being made to make it less noticeable.

[Problem to be solved by the invention]

However, according to the former method, since the DC component is multidimensionally distributed on the image, the amount of interpolation becomes large, and there is a risk that original information may be lost due to correction, which is a clinical problem. Furthermore, the latter has the problem that some DC noise may remain in the center of the reconstructed image. This is mainly caused by time fluctuations in the zero level of the amplifier in the NMR signal detection circuit system. According to the latter - Mx M (128x 128.2
In MXM image reconstruction when collecting matrix data (such as 56x 256), the noise at the center is only one point, and there is no particular problem with pixel interpolation after reconstruction, but in various 0 padding reconstructions, etc., this one point noise Because the image spreads in multiple dimensions, the reliability of image diagnosis becomes unreliable. This 0-padding reconstruction method is used for phase encode reduction to shorten scan time and double matrix (2MX2M) reconstruction to obtain high resolution, and is clinically important. If there is a problem that the reliability of image diagnosis is lost due to noise, this zero padding reconstruction method cannot be used effectively. An object of the present invention is to provide an improved MHI device that effectively removes DC component noise without losing original information.

[Means to solve the problem]

In order to achieve the above object, the MHI device according to the present invention includes a multidimensional Fourier transform means for performing multidimensional Fourier transform on data collected by excitation with a high frequency signal and a pulse sequence of NMR signal reception, and a multidimensional Fourier transform means that does not perform excitation with a high frequency signal. means for detecting a DC component by performing at least one-dimensional Fourier transform on the data obtained by executing the above-mentioned pulse sequence; The method is characterized in that it includes means for subtracting from the data.

[For production]

If data is collected by executing a pulse sequence without excitation with a high-frequency signal, data consisting only of reception system noise will be collected. Therefore, after performing at least one-dimensional Fourier transform on this data, the DC component as the data after the at least one-dimensional Fourier transform is detected. Raw data of the NMR signal is obtained by a normal pulse sequence, and after this is subjected to at least one-dimensional Fourier transform, the above-mentioned DC component is subtracted. Then, the data obtained by at least one-dimensional Fourier transformation will have DC component noise removed. By performing two-dimensional and subsequent Fourier transforms on the data after at least one-dimensional Fourier transform that has been corrected in this way, an image from which DC component noise has been removed is reconstructed.

【Example】

Hereinafter, an MHI device according to an embodiment of the present invention will be described in detail with reference to the drawings. In Figure 1,
The waveform generation circuit 11 changes the spin magnetization from 90° to 180°.
A tilted amplitude modulation waveform is generated, converted into an analog signal by the D/A converter 12, and then sent to the amplitude modulation circuit 13. In the amplitude modulation circuit 13, the carrier wave (high frequency) signal sent from the generation circuit 16 is amplitude modulated with the above amplitude modulation waveform, and the modulated high frequency signal is sent to the transmitting coil 15 via the power amplifier 14. , an excitation is performed on a test object such as a human body (not shown).The NMR signal (echo signal) generated in the test object such as a human body is received by the receiving coil 17, detected by the detection circuit 18, and then passed through the low-pass filter 19 to the A. /D converter 20 samples and converts it into a digital signal.This digital signal is taken into the host computer 21 and subjected to multidimensional (two-dimensional) fast Fourier transform (FF
T) etc. are processed. Next, an example of processing when double matrix reconstruction is performed by a zero padding reconstruction method using a two-dimensional Fourier transform method will be described with reference to a flowchart shown in FIG. First, in step 31, a Buri sequence is performed, and the noise N(
Collect t). That is, at this time, the carrier wave from the carrier wave generating circuit 16 is interrupted by a command from the host computer 21, and in this state, a pulse sequence of excitation and reception by a high frequency signal is performed. Here, since the carrier wave that provides the resonant frequency that generates the NMR phenomenon is blocked, no echo signal is generated at all, and the host computer 21 receives data N (t ) are imported. In this embodiment, this N (t) is the A/D converter 20''
This was obtained by sampling the rM point, as shown in FIG. In step 32, this N(t) minus the DC component is subjected to one-dimensional Fourier transformation. Here, the DC component is roughly detected by, for example, taking the average of the data N(t), so the data n(t) obtained by subtracting this component is as shown in Figure 3. It contains some direct current components and adjacent low frequency components. Since this n(t) consists of M pieces of data obtained by M-point sampling, 0 padding is performed at both ends of each M/2 point to obtain 2M pieces of data, and then a one-dimensional Fourier transform is performed, and n(ω ). Since n(t) is mostly white noise, n(
ω) is −-like noise, but due to the accuracy of the coarse detection of the DC component described above, errors occur in the DC and adjacent low frequency components as shown in FIG. Therefore, in step 33, from this n(ω), the DC Po
and the absolute values of its adjacent points P-1 and P+1 are determined and stored. Next, in step 34, after recovering the carrier wave from the carrier wave generation circuit 16, while changing the amount of phase encoding, normal high frequency signal excitation/NMR
By repeating the reception pulse sequence, raw data S (t, phs) of the MXM matrix within the small frame A in Fig. 4 is obtained.
Collect. Then, in step 35, the rough detected DC component described above is subtracted from the raw data S (t, phs), and then zero padding is performed at M/2 points at both ends in the frequency direction to obtain 2M pieces of data. A dimensional Fourier transform is performed to obtain S(ωX, ρhs) as shown by B in FIG. Next, in step 36, S (
The above P-1, Po, and p+i components are subtracted from (ωx, phs). As an example, to explain the case of subtracting the Po component from the center (DC) data of S(ωX, ρhs), first transform S(ωx, phs) into polar coordinates as shown in Figure 5 and convert it to (r, θ). seek. And r=r-P. , and S in the orthogonal coordinate system corresponding to (r゛, θ)
' Find (ωx, phs>. DC adjacent point P-1,
Similarly for P+1, S' (ωx, phs)
seek. This allows S(ωx
, phs), corrected data S' (ωχ, phs) can be obtained by removing the direct current and its adjacent low frequency components. Do this for all phs lines and s'
(ωχ, phs) is obtained. Next, in step 37, M/2 is applied to both ends in the phase direction for the above corrected data S' (ωx, phs).
Perform line zero padding to obtain 2MX2M matrix data, perform second-dimensional Fourier transform (Fourier transform in the phase direction), and then take the absolute value of the complex number in step 38 to obtain a two-dimensional image. For example, an image of a 2MX×2M matrix from which direct current and adjacent low frequency components have been completely removed, as shown by C in FIG. 4, can be obtained by the double matrix reconstruction method. Since the above-described correction processing can be performed during scanning, the image reconstruction time is not substantially extended compared to the case where such correction is not performed. Note that although the above embodiment uses a double matrix reconstruction method based on a two-dimensional Fourier transform method, in the case of a three-dimensional Fourier transform method, it is sufficient to similarly apply compensation to intermediate data after one-dimensional Fourier transform. It goes without saying that it can also be applied to techniques such as encode reduction.

【Effect of the invention】

According to the MHI device of the present invention, DC component noise and adjacent low frequency component noise can be effectively removed,
A highly reliable image can be reconstructed. and 0
Since image reliability is not lost even with the padding reconstruction method, the imaging time can be shortened by adopting the padding reconstruction method, or high-resolution images can be obtained without extending the imaging time. can.

[Brief explanation of drawings]

Fig. 1 is a block diagram of an embodiment of the present invention, Fig. 2 is a flowchart for explaining the operation, Fig. 3 is a waveform diagram showing each waveform in the pre-sequence, and Fig. 4 is a raw data and
A diagram showing data after dimensional Fourier transformation and a reconstructed image,
FIG. 5 is a diagram for explaining PO acid component subtraction. 11... Waveform generation circuit, 12... D/A converter, 13... Amplitude modulation circuit, 14... Power amplifier,
15... Transmission coil, 16... Carrier wave generation circuit, 1
7... Receiving coil, 18... Detection circuit, 19...
Low-pass filter, 20... A/D converter, 21
...Host computer.

Claims (1)

[Claims] (1) A multidimensional Fourier transform means for performing multidimensional Fourier transform on data collected by a pulse sequence of excitation with a high frequency signal and reception of an NMR signal, and data obtained by executing the above pulse sequence without excitation with a high frequency signal. The present invention is characterized by comprising means for detecting a DC component by performing at least one-dimensional Fourier transform on the data, and means for subtracting this DC component from data that has been subjected to at least one-dimensional Fourier transform in the multidimensional Fourier transform means. MRI machine.