Magnetic resonance imaging apparatus

[Technical field to which the invention belongs] The present invention measures a nuclear magnetic resonance (hereinafter referred to as "NMR") signal from hydrogen, phosphorus, etc. in a subject, and obtains a nuclear density distribution, relaxation time distribution, and the like. Regarding the nuclear magnetic resonance imaging (MRI) device to be visualized, in areas such as the heart where high-speed imaging is required, in particular, multiple RF receiving coils are used, and the signal acquired by thinning out the phase encoding in each RF receiving coil is RF. The present invention relates to an MRI apparatus capable of obtaining an image without artifacts in an imaging method (hereinafter referred to as a parallel imaging method) developed by matrix calculation using the sensitivity distribution of a receiving coil.

PROBLEM TO BE SOLVED: To obtain an echo signal necessary for reconstructing one image by repeatedly executing a sequence while changing the amount of phase encoding in MRI. Therefore, the number of repetitions greatly affects the image acquisition time. When performing high-speed photography, generally, a multi-echo type sequence that generates multiple echo signals within one repetition is used, or a sequence that shortens the repetition time interval to several to several tens of ms is used. .. However, the multi-echo type sequence may reduce the contrast of the image or cause image distortion. This causes a decrease in contrast because the echo time that contributes to the contrast of the image is different for each echo signal in the multi-echo type sequence. Further, since the echo time is different, if the phase change between the echo signals is different, the image is distorted and appears in the image.

[0003] Further, when photographing a cardiac region (such as coronary artery imaging), it is necessary to acquire an image at a higher speed, and a high-speed imaging method called a parallel imaging method has also been proposed. In the parallel imaging method, a multiple RF receiving coil is used to execute a sequence in which phase encoding is thinned out at equal intervals to reduce the number of repetitions and shorten the imaging time. Normally, when the phase encoding is thinned out at equal intervals and measurement is performed, wrapping occurs in the image, but the image is expanded and the wrapping is removed by performing matrix calculation based on the sensitivity distribution of each RF receiving coil. .. Generally, in the parallel imaging method, the imaging time can be shortened by the number of RF receiving coils used for imaging.

[0004] In the parallel imaging method, the shooting time can be shortened without using a multi-echo type sequence, so that the effects of contrast reduction and image distortion can be reduced. .. However, the resulting image changes greatly depending on the arrangement of the multiple RF receiving coils and the shape of the sensitivity distribution. In particular, if a matrix operation is performed while including a low signal region such as a background, an error becomes large at the time of unfolding due to the influence of noise, and an artifact of a bright spot may occur in the image. In addition, in order to calculate the sensitivity distribution of each RF receiving coil with high accuracy, it is desirable to calculate the sensitivity distribution using an image measured with a body coil for the whole body with a relatively uniform sensitivity distribution, but depending on the device, it is for the whole body. There is a problem that images cannot be acquired at the same time because there is no body coil or the number of receiving channels is small.

[0005] Therefore, an object of the present invention is to suppress a calculation error during matrix calculation, eliminate the occurrence of bright spot artifacts, and enable parallel imaging even in a device having no body coil for the whole body or a device having a small number of channels. There is. Alternatively, even in a device having a body coil for the whole body, the device configuration and the signal processing flow are to be simplified.

[Means for Solving the Problems] In order to achieve the above object, the present invention includes a receiving means for receiving a nuclear magnetic resonance signal from a subject provided with a plurality of RF receiving coils, and the receiving means. The nuclear magnetic resonance signal is measured by thinning out the k-space encoding steps using the above, and the measurement control means for acquiring the k-space data for each RF receiving coil and the sensitivity image for each RF receiving coil are used. Using the sensitivity distribution calculation to acquire the sensitivity distribution data for each RF receiving coil, and the k-space data and sensitivity distribution data for each RF receiving coil.<u style="single">Based on parallel imaging method</u>In a magnetic resonance imaging apparatus including an image reconstruction operation for reconstructing an image of a subject by applying an operation and a calculation means for performing the calculation, the sensitivity distribution calculation is performed.<u style="single">At least two sensitivity images of the plurality of RF receiving coils are combined to obtain a substantially uniform sensitivity image, and the substantially uniform sensitivity image is used to at least one RF receiving coil of the plurality of RF receiving coils. Get the sensitivity distribution data of</u>It is characterized by that.

[0007] Further, at least two RF receiving coils are provided, and the measurement of the nuclear magnetic resonance signal received by the RF receiving coil is measured so as to thin out the encoding steps in the measurement space, and the sensitivity image and the form are measured for each RF receiving coil. Magnetic resonance imaging having a control means for acquiring an image, performing an operation of folding back removal of each morphological image from a sensitivity distribution based on the sensitivity image of the RF receiving coil, and combining the morphological images to obtain one morphological image. In the apparatus, the control means creates an overall sensitivity image by combining the sensitivity images acquired by each RF receiving coil, calculates the sensitivity distribution of each RF receiving coil using the overall sensitivity image, and obtains the overall sensitivity image. A mask that separates a non-image area and an image area of an image is created by using the mask, and an operation is performed based on the mask.

[0008] Further, a first RF receiving coil for receiving the entire measurement area and a plurality of second RF receiving coils for receiving a region obtained by dividing the measurement area by at least two or more are provided, and each of the second RFs is provided. The measurement of the nuclear magnetic resonance signal received by the receiving coil is measured so as to thin out the encoding steps in the measurement space, a sensitivity image and a morphological image are acquired for each of the second RF receiving coils, and the sensitivity by the first RF receiving coil is obtained. Magnetic resonance having a control means for performing an operation of folding back removal of the morphological image from the image and a sensitivity distribution obtained from the sensitivity image obtained by the second RF receiving coil and combining the morphological images to obtain one morphological image. In the imaging apparatus, the control means creates a mask in which a non-image region and an image region are separated in the reception sensitivity distribution, and performs an operation using the mask.

BEST MODE FOR CARRYING OUT THE INVENTION The nuclear magnetic resonance imaging apparatus of the present invention will be described in detail below with reference to the drawings. Figure 4 shows the configuration of a typical nuclear magnetic resonance imaging device. A magnet 402 that generates a static magnetic field around the subject 401, a gradient magnetic field coil 403 that generates a gradient magnetic field in the space, an RF coil 404 that generates a high-frequency magnetic field in this region, and an MR signal generated by the subject 401. There is an RF probe 405 to detect. The gradient magnetic field coil 403 is composed of gradient magnetic field coils in three directions of X, Y, and Z, and generates a gradient magnetic field in response to a signal from the gradient magnetic field power supply 409. The RF coil 404 generates a high frequency magnetic field in response to the signal of the RF transmitter 410. The signal of the RF probe 405 is detected by the signal detection unit 406, processed by the signal processing unit 407, and converted into an image signal by calculation. The image is displayed on the display unit 408. The gradient magnetic field power supply 409, RF transmitter 410, and signal detection unit 406 are controlled by the control unit 411, and the control time chart is generally called a pulse sequence. Bed 412 is for the subject to lie down.

[0010] Currently, the objects to be imaged by MRI are protons, which are the main constituents of the subject, as those that are widely used clinically. By imaging the spatial distribution of the proton density and the spatial distribution of the relaxation phenomenon of the excited state, the morphology or function of the human head, abdomen, limbs, etc. can be photographed two-dimensionally or three-dimensionally.

[0011] As an example of the RF coil 405, a technique called "multiple RF coil" or "phased array coil" using a plurality of receiving coils is used. Multiple RF coils are made by arranging multiple small RF receiving coils with relatively high sensitivity and synthesizing the signals acquired by each coil to expand the field of view while maintaining the high sensitivity of the RF receiving coils, resulting in high sensitivity. This is a reception-only RF coil. As a multiple RF coil for the horizontal magnetic field head, there is Array Head Coil for Improved Functional MRI (Christoph Leussler), 1996 ISMRM abstruct p.249. Helmet and Cylindrical Shaped CP Array Coils for Brain Imaging: There is A Comparison of Signal-to-Noise Characteristics (HAStark, EMHaacke), 1996 ISMRM abstract p.1412. Four Channel Wrap-Around Coil with Inductive Decoupler for 1.5T Body Imaging (T. Takahashi et.al), 1995 ISMRM abstruct p.1418 are examples of QD multiple RF coils for the abdomen of horizontal magnetic field.

[0012] A part of the signal detection unit of the multiple RF coil is shown in FIG. In FIG. 3, four RF receiving coils 405 are each connected to the preamplifier 302 to form one multiple coil 301. In the signal detection unit 406, four AD conversion / orthogonal detection circuits 303 are arranged in parallel, and the outputs of the respective preamplifiers are connected to each other. The signal processing unit 407 is composed of an operation 304 that obtains an MRI image detected by each RF receiving coil 405 by Fourier transform, back projection method, wavelet transform, etc., and synthesizes these image signals.

[0013] Next, a shooting method will be described. Figure 6 shows a typical gradient echo sequence. 601 is a high-frequency pulse, 602 is a slice-selective gradient magnetic field pulse, 603 is a phase-encoded gradient magnetic field pulse, 604 is a readout gradient magnetic field pulse, 605 is a sampling window, 606 is an echo signal, and 607 is the repetition time (irradiation interval of the high-frequency pulse 601). Is. In MRI, the amount of the phase-encoded gradient magnetic field pulse 603 is changed for each repetition time 607, different phase encodings are given, and the echo signal 606 obtained by each phase encoding is detected. This operation is repeated for the number of phase encodings, and the echo signal required for reconstructing one image is acquired at the image acquisition time of 608. The number of phase encodings is usually selected to be 64, 128, 256, 512, etc. per image. Each echo signal is usually obtained as a time-series signal consisting of 128, 256, 512, and 1024 sampled data. Two-dimensional Fourier transform is performed on these data to create one MR image.

[0014] In the case of the parallel imaging method, which is high-speed imaging using a multiple RF receiving coil, the phase encoding step interval is thinned out at a constant rate to reduce the number of repetitions of imaging. This thinning rate is generally called a double speed number. For example, if the phase encoding step is thinned out twice, the multiple speed number is 2. Hereinafter, the principle of the parallel imaging method will be described with reference to FIG. In the case of normal shooting, as shown in Fig. 2 (a), the signals 202n acquired at each phase encoding amount are arranged to form data 201 for one image, which is Fourier transformed to show Fig. 2 (c). Get an image like. Next, in the parallel imaging method, for example, when the phase encoding step interval is doubled and the data is thinned out, as shown in FIG. 2 (b), the data 204 m is measured every other line and corresponds to the position of 205 m. No data is measured. At this time, since the amount of measured data 204 m is halved compared to normal shooting, the matrix is halved to create an image, but an image with wrapping as shown in FIG. 2 (d) can be obtained. In FIG. 2, the y direction is the phase encoding direction, and this folding occurs when the image in the phase encoding direction is folded. That is, the subject image 2061 in the upper region 2071 and the subject image 2062 in the lower region 2072 of the subject 206 in the image 207 as shown in FIG. 2 (c) overlap, and are shown in FIG. 2 (d). The image 208 has such a wrap. The signal wrapping generated in this way is removed by a signal processing method called SENSE (SENSE: Sensitivity Encoding for Fast MRI (Klaas P. Pruessmann et.al), Magnetic Resonance in Medicine 42: 952-962 (1999)). ).

[0015] Hereinafter, folding back removal will be described. When the image matrix in the x and y directions is X and Y, respectively, the pixel value at the coordinates (x, y) (x: 1 x X, y: 1 y Y) in the image is s._i Let (x, y) (where i is the coil number and 2 i N). In the case of FIG. 2D, since the phase encoding steps are thinned out twice, the matrix in the phase encoding direction of the image after thinning out is Y' Y / 2. When the coordinates of the image in FIG. 2 (d) are (x, y') (1 y' Y'), the pixel value s'(x, y') 208 is the two regions 2071 of the original image 207. , 2072 overlap, [0016] [number 1]<img file="JP4047553B2_D0001.tif" />Will be. Where a represents a constant. Next, the sensitivity distribution and the image of the RF receiving coil will be described. The two-dimensional sensitivity distribution of the i-th RF receiving coil, c_iIf (x, y), the received signal s_i (x, y) is the sensitivity distribution c of the receiving coil_iThe product of (x, y) and the proton density distribution p (x, y) of the subject, [0017] [Equation 2]<img file="JP4047553B2_D0002.tif" />It is represented by. Using equation (2), equation (1) is [0018] [Equation 3].<img file="JP4047553B2_D0003.tif" />[0019] [Number 4]<img file="JP4047553B2_D0004.tif" />Then, equation (3) is [0020] [number 5].<img file="JP4047553B2_D0005.tif" />Will be. This can be represented as an N-by-2 matrix, [0021] [Equation 6].<img file="JP4047553B2_D0006.tif" />Will be. From this, the sensitivity distribution C of the coil_ijIf is known, the proton density distribution P of the subject can be calculated by calculating the inverse matrix._jI understand.

Similarly, when shooting is performed at M times speed using N coils, Y' Y / M and 1 y'Y'are set as [0023] [Equation 7].<img file="JP4047553B2_D0007.tif" />Will be. Where b represents a constant. The relationship between the number of coils and the number of double speeds in the parallel imaging method is mathematically N M because the wrapping is removed by the inverse matrix operation.

[0024] Normally, in the parallel imaging method, the sensitivity distribution C of each RF receiving coil is used._ijIs acquired by measuring in advance by pre-measurement or the like. However, the sensitivity distribution C_ijIs difficult to calculate directly, and in general, using an image of a body coil for the whole body with a relatively uniform sensitivity distribution, divide the image acquired by each RF receiving coil to obtain an approximate coil sensitivity distribution. Find and perform matrix operation.

FIG. 7 shows the processing of a general parallel imaging method. The figure shows the configuration of one body coil for the whole body and three multiple RF receiving coils. First, the sensitivity distribution calculation processing 7041 to 7043 is performed using the sensitivity images 701 acquired by the body coil for the whole body and the sensitivity images 7021 to 7023 acquired by each RF receiving coil, and the sensitivity distribution 7051 to 7053 of each RF receiving coil is performed. To get. As the sensitivity distribution calculation process 704, the image of the body coil for the whole body is s._cAs (x, y), for example, [0026] [Equation 8]<img file="JP4047553B2_D0008.tif" />There is. Using the sensitivity distributions 7051 to 7053 calculated in this way and the measured images 7031 to 7033 with wrapping acquired by each RF receiving coil, a matrix is created by the matrix creation process 706 and then folded by the inverse matrix calculation process 707. The image 708 from which is removed is obtained. Next, the first embodiment of the present invention will be described with reference to FIGS. 1 and 5. In the conventional general parallel imaging method, no consideration is given to the low signal region such as the background, but in the present embodiment, the low signal region is taken into consideration. That is, using the image of the body coil for the whole body, the mask 102 was created by the mask creation step 101, and the processed mask 104 with improved accuracy was created by the mask processing step 103, and the mask was used. It is configured to perform the matrix creation process 105. Here, the mask processing will be described. Figure 5 shows the case of triple speed, and image 501 is the result of overlapping the three regions 5041, 5042, and 5043 of the normal image (Fig. 5 (b)) (where Y' Y / 3, 1 y). 'Y'). Two points of interest A (x) in Fig. 5 (a)_A, y'_A) 502, B (x_B, y'_B) Provide 503. At this time, since the pixel values of points A and B overlap the regions 5021 to 5023 and 5031 to 5033 in the figure, respectively, [0027] [Equation 9]<img file="JP4047553B2_D0009.tif" />Will be. Where d represents a constant. However, 5021 is a signal of the background part without a subject, and when a matrix operation is performed using this, the matrix may diverge during the inverse matrix operation, and an artifact of a bright spot may occur in the result image. .. Therefore, in order to eliminate the influence of such a background, the mask m (x, y) 102 as shown in FIG. 5 (c) is used. As the mask creation process 101 in FIG. 1, for example, a threshold value is set for the sensitivity image 701 of the body coil for the whole body, and pixels having a pixel value equal to or less than the threshold value as shown in FIG. The pixel whose value is equal to or greater than the threshold value is defined as the subject area 505. Further, as a threshold value setting method, for example, about 1/10 of the maximum pixel value in the image is set as the threshold value, or a pixel value histogram 507 in the image as shown in FIG. 5 (d) is created to set the noise distribution. It is also possible to obtain and set the threshold value 508. Next, the mask creation 101 obtains a mask 102 in which, for example, the value of the subject area 505 is set to 1 and the value of the background area 506 is set to 0. Then, the obtained mask 102 is subjected to mask processing 103 in order to further improve the accuracy of the mask. Then, the matrix creation process 105 is performed using the processed mask 104 obtained by the mask process 103. As a result, the matrix operation can be performed excluding the background area, so that the occurrence of bright spot artifacts can be eliminated.

[0028] Here, as the mask processing 103, for example, when a structure exists inside the subject, the signal of that region is used as it is, so that the mask region is filled by interpolation or the region that is not originally included in the calculation is mistakenly used. Perform processing to remove the isolated points that have been taken. Further, in the matrix creation process 105 using the mask, a discrimination process is performed in which each element is set to 0 so that the background area is not included in the matrix elements. For example, [0029] [number 10]<img file="JP4047553B2_D0010.tif" />It is represented by. Further, when the background area is set to 0 and the image becomes unnatural, the matrix elements may be set to a predetermined constant for processing. For example, [0030] [Number 11]<img file="JP4047553B2_D0011.tif" />It is represented by. Here, Const. Indicates an arbitrary constant. Furthermore, using the discrimination process in which each element is set to 0 as shown in Eq. (10) in the matrix creation process 105 is the product of the sensitivity distribution and the mask, [0031] [Equation 12].<img file="JP4047553B2_D0012.tif" />Therefore, the discrimination process can be eliminated and the process can be simplified. Setting a part of the matrix element to "0" as in Eq. (10) is equivalent to reducing the matrix size of the matrix. That is, "the situation of folding is examined for each pixel, the necessary and sufficient matrix size is determined, and the necessary and sufficient calculation is performed". As a result, as compared with the conventional case, unnecessary noise is eliminated and the image quality is improved. There is also an advantage that the calculation time is shortened.

[0032] In the matrix creation process 105, the calculation is performed using the mask 104 after the processing, but the calculation can also be performed using the mask 102. This makes it possible to simplify the processing flow. Further, although the mask 102 has been described separately for the background region and the subject region, the background region includes not only the background outside the subject region but also the low signal region (for example, voids and the like) existing in the subject region.

Next, a second embodiment of the present invention will be described with reference to FIG. In this case, there is no image of the body coil for the whole body. In this case, the image of the body coil for the whole body necessary for calculating the sensitivity distribution of each RF receiving coil is created by the signal coupling process 801. As a specific process of 801 for example, the image s acquired by each RF receiving coil._i(x, y) and s_iImage with low-pass filter applied to (x, y) w_iWhen (x, y), the pseudo body coil image after composition s'_c(x, y) is [0034] [number 13]<img file="JP4047553B2_D0013.tif" />(Here, * represents the complex conjugate and || represents the absolute value). However, since the sensitive region of the multiple RF receiving coil is narrower than that of a normal coil, the image of the pseudo body coil synthesized in this way may be greatly affected by shading. Therefore, the sensitivity correction process 802 is applied to the synthesized pseudo body coil image to create a pseudo image 803 without shading, and the parallel imaging method is performed. As the sensitivity correction process 802, for example, a method of calculating and correcting shading from an image (Japanese Patent Laid-Open No. 7-222724) is used.

By synthesizing the signals of the body coil for the whole body in this way, for example, the parallel imaging method can be performed even in a device having no body coil for the whole body or cannot be attached, or a device having a small number of usable coil channels. Is possible. Further, even in a device having a body coil for the whole body, only the signal of the multiple RF receiving coil is processed, so that the device configuration, software configuration, and data flow can be simplified. For example, in a hamburger type open MRI, the sensitivity of the irradiation coil is relatively low, so that reception by the irradiation coil may not be possible. Moreover, even if it can be physically received, if a receiving component is attached only for the parallel imaging method, the cost of the MRI apparatus will increase. However, by applying this embodiment, such disadvantages can be overcome. The calculation of the sensitivity distribution can also be applied to multi-slice photography and three-dimensional photography. In multi-slice, equation (13) is calculated for each slice. Even in 3D photography, equation (13) is calculated for each slice.

[0036] The present invention is not limited to the contents disclosed in the above examples, and may take various forms based on the gist of the present invention. In this embodiment, an example of a parallel imaging method using three multiple RF receiving coils is shown, but the number N of RF receiving coils can be any number of 2 or more. Moreover, although the example of the double speed number M = 2 or 3 is shown, the double speed number can be selected within the range of M N. Further, although the gradient echo sequence has been described in the present invention, the parallel imaging method does not depend on the shape of the sequence. For example, it can be applied to SE sequence, FSE sequence, EPI sequence, spiral sequence, SSFP sequence and the like. Further, when the present invention is applied to three-dimensional measurement, data can be thinned out not only in the phase encoding direction but also in the slice encoding direction to increase the speed. Alternatively, the data can be thinned out by combining the phase encoding direction and the slicing direction to increase the speed. Furthermore, when cardiac imaging is performed by applying this algorithm, an image without bright spot artifacts can be acquired with high time resolution.

[Effect of the Invention] Since the present invention is configured as described above, the bright spot artifact can be eliminated by the parallel imaging method. In addition, the parallel imaging method can be performed even in a device without a body coil for the whole body. Further, even in a device having a body coil for the whole body, the parallel imaging method can be performed only by the signal of the multiple RF receiving coil, so that the number of channels used can be reduced and the device configuration, processing, and data flow can be simplified.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram illustrating signal processing according to the first embodiment of the present invention.

FIG. 2 is a diagram illustrating image wrapping in the parallel imaging method.

FIG. 3 is a diagram showing a receiving unit of an RF coil to which the present invention is applied.

FIG. 4 is a diagram showing an overall configuration of an MRI apparatus to which the present invention is applied.

FIG. 5 is a diagram illustrating folding of an image in the present invention.

FIG. 6 is a diagram illustrating a general gradient echo sequence.

FIG. 7 is a diagram illustrating a process of a conventional parallel imaging method.

FIG. 8 is a diagram illustrating signal processing according to a second embodiment of the present invention.

[Code description] 601 high frequency pulse, 602 slice selection gradient magnetic field, 603 phase encode gradient magnetic field pulse, 604 readout gradient magnetic field pulse, 605 data sample window, 606 echo signal, 607 repetition time interval, 608 image acquisition time, 401 subject , 402 magnet, 403 gradient magnetic field coil, 404 RF coil, 405 RF probe, 406 signal detector, 407 signal processing unit, 408 display unit, 409 gradient magnetic field power supply, 410 RF transmitter, 411 control unit, 412 beds