JP5819675B2 - Magnetic resonance imaging apparatus and program - Google Patents

Description

  The present invention relates to a magnetic resonance imaging apparatus and a program thereof. In particular, the present invention relates to a magnetic resonance imaging apparatus that complements unsampled data that has not been sampled from reference data by a technique such as convolution integration, and a program thereof.

  In order to shorten the imaging time of the subject, various ideas have been made for the magnetic resonance imaging apparatus. For example, Patent Document 1 discloses a method of parallel imaging. In the parallel imaging method, the nuclear magnetic resonance signal from the subject is detected by thinning out the encoding step in the phase direction. As a result, the parallel imaging technique performs imaging by shortening the imaging time, and generates an image in which the imaging field of view narrows and is folded back and a false image (artifact) exists. Then, the parallel imaging technique turns back the false image based on the difference in sensitivity distribution of the plurality of coils, and finally obtains an image with a wide field of view. Parallel imaging techniques can reduce scan time, reduce image blur and geometric distortion, improve spatial or temporal resolution, and expand volume coverage.

  Japanese Patent Application Laid-Open No. 2004-228561 discloses a method in which parallel imaging and compressed sensing are combined. Compressed sensing is a technique that takes advantage of the fact that most medical images have some degree of “compressibility” and sets a significant number of values to zero with little loss in image quality. Compressed sensing can reduce scan time, image blur and geometric distortion. Both parallel imaging techniques and compressed sensing techniques can reconstruct undersampled raw data (raw data). Here, undersampling is a technique for sampling shooting data in which a part of shooting data in k-space is missing (unsampled).

JP 2004-344327 A JP 2009-268901 A

  However, sparse k-space shooting data in which there is no part of k-space shooting data is unsampled data that is not sampled. Since the image is reconstructed based on the sparse k-space image data, both the parallel imaging method and the compression sensing method may deteriorate the image quality. Therefore, an image reconstruction technique that can acquire a high-quality image in a short scanning time is desired.

  A magnetic resonance imaging apparatus according to a first aspect includes a transmission coil that generates a high-frequency magnetic field pulse with respect to a subject placed in a static magnetic field, a signal detection unit that detects a nuclear magnetic resonance signal generated from the subject, Is provided. The magnetic resonance imaging apparatus generates a first pulse sequence generated from a transmission coil to sample k-space reference data and an under-sampling data lacking a part of k-space data to generate from a transmission coil. Acquisition of a correlation function between the reference data and the undersampling data based on the second pulse sequence thus performed, the reference data obtained in the first pulse sequence, and the undersampling data obtained in the second pulse sequence Means, and image reconstruction means for reconstructing image data of a portion lacking k-space based on the correlation function.

  A magnetic resonance imaging apparatus according to a second aspect is the magnetic resonance imaging apparatus according to the first aspect, wherein the first pulse sequence includes a pulse sequence for obtaining a proton density image as reference data, and the second pulse sequence is an underscore. The sampling sequence includes a pulse sequence for obtaining a T1-weighted image, a T2-weighted image, an in-phase image, or an out-of-phase image.

  A magnetic resonance imaging apparatus according to a third aspect is the magnetic resonance imaging apparatus according to the first aspect, wherein the signal detection means outputs a nuclear magnetic resonance signal generated by the first pulse sequence to the first reception coil and the second reception coil. The signal detecting means receives the nuclear magnetic resonance signal generated by the second pulse sequence by the first receiving coil and the second receiving coil.

  A magnetic resonance imaging apparatus according to a fourth aspect is the magnetic resonance imaging apparatus according to the third aspect, wherein the function acquisition means uses reference data and undersampling data obtained by one of the first reception coil and the second reception coil. Based on the above, one correlation function is obtained, and the image reconstruction means uses one correlation function for image reconstruction of the other data of the first reception coil or the second reception coil.

  A magnetic resonance imaging apparatus according to a fifth aspect is the magnetic resonance imaging apparatus according to any one of the first to fourth aspects, wherein the function acquisition means includes a central region of the k space in the undersampling data. A correlation function between the reference data and the undersampling data is acquired based on the undersampling data.

  A magnetic resonance imaging apparatus according to a sixth aspect is the magnetic resonance imaging apparatus according to any one of the first to fourth aspects, wherein the function acquisition means includes the amplitude of the k space in the undersampling data. A correlation function between reference data and undersampling data is acquired based on undersampling data in a large area.

A magnetic resonance imaging apparatus according to a seventh aspect is the magnetic resonance imaging apparatus according to any one of the first to sixth aspects, wherein the k-space reference data is REF (kx, ky), and the k-space When the undersampling data is CON (kx, ky), the kernel (integral kernel) F (kx, ky) satisfying the following formula 1 is a correlation function.
CON (kx, ky) = F (kx, ky) * REF (kx, ky)...
However, * is convolution (convolution integration).

  The magnetic resonance imaging apparatus according to an eighth aspect is the magnetic resonance imaging apparatus according to any one of the first to seventh aspects, wherein the image reconstruction unit has a spatial dependence of the sensitivity of the signal detection unit. Parallel imaging for improving the space or temporal resolution of the image is used to reconstruct the data of the portion lacking k-space.

  The program according to the ninth aspect is a program for obtaining a high-quality image by shortening the pulse sequence time when an object is imaged by the magnetic resonance imaging apparatus. The program causes a computer used in the magnetic resonance imaging apparatus to generate a first pulse sequence for sampling k-space reference data, and a second to sample undersampling data lacking some data in k-space. A procedure for generating a pulse sequence, a function acquisition procedure for acquiring a correlation function between the reference data and the undersampling data based on the reference data and the undersampling data, and a thinning of the undersampling data based on the correlation function And an image reconstruction procedure for reconstructing the image of the data at the selected location.

  The magnetic resonance imaging apparatus of the present invention calculates unsampled k-space imaging data from a correlation function between reference data and undersampling imaging data. Since unsampled k-space imaging data can be interpolated, the magnetic resonance imaging apparatus can acquire a high-quality image based on the interpolated k-space imaging data.

1 is a schematic configuration diagram of a magnetic resonance imaging apparatus 10. FIG. (A) is an example of the imaging data of k space in a predetermined part. (B) is an example of the probability density distribution on the kx line of the probability density function. It is a figure which shows the area | region where a probability becomes larger than threshold value TH in a probability density function. It is the flowchart which showed the acquisition procedure of contrast image CI2. (A) is a schematic diagram of the k space of the reference image RI arranged two-dimensionally. (B) is a schematic diagram of the reconstructed reference image RI. (A) is a schematic diagram of the k space of the contrast image CI1 two-dimensionally arranged. (B) is a schematic diagram of a contrast image CI1 reconstructed. It is the figure which showed the relational expression of Numerical formula 2 in the image space. FIG. 6 is a schematic diagram of imaging data ci1 of a contrast image CI1 undersampled in k space, imaging data ri of a fully sampled reference image RI, and a correlation function f1 (kx, ky). (A) is a schematic diagram of unsampled shooting data d (x, y) in the k space of the contrast image CI1, a correlation function F (kx, ky), and shooting data REF (kx, ky) of the reference image RI. is there. (B) is a diagram showing dense photographing data ci2. (C) is a diagram showing a contrast image CI2 that has been subjected to Fourier transform FT using dense photographing data ci2. (A) is a figure showing an example of multi-receiver 80. (B) is the figure which showed the 1st receiving coil group. (C) is the figure which showed the 2nd receiving coil group.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. The scope of the present invention is not limited to these forms.
<Configuration of magnetic resonance imaging apparatus>

  FIG. 1 is a schematic configuration diagram of a magnetic resonance imaging apparatus 10 of the present embodiment. With reference to FIG. 1, the configuration and basic operation of the magnetic resonance imaging apparatus 10 of the present embodiment will be described.

  The magnetic resonance imaging apparatus 10 of the present embodiment includes a magnet system 100, a gradient coil drive unit 130, an RF coil drive unit 140, a data collection unit 150, a sequence control unit 160, a data processing unit 170, a display unit 180, and an operation unit 190. Have.

  The magnet system 100 includes a main magnetic field coil unit 102, a gradient coil unit 106, and an RF coil unit 108. Each of these coil portions has a substantially cylindrical shape, and is arranged coaxially with each other in a substantially columnar bore. The subject SB is placed on the bed 200 in the bore, and the bed 200 can move in the bore in the magnet system 100 according to the imaging region.

  The main magnetic field coil unit 102 forms a static magnetic field in the internal space of the magnet system 100. The direction of the static magnetic field is generally parallel to the direction of the body axis of the subject SB and forms a horizontal magnetic field. The main magnetic field coil unit 102 is normally configured using a superconducting coil, but may be configured using a permanent magnet or the like without being limited to the superconducting coil.

  The gradient coil unit 106 has three types of gradients for imparting gradients to the static magnetic field strength formed by the main magnetic field coil unit 102 in the three axes orthogonal to each other, that is, in the direction of the slice axis, the phase axis, and the frequency axis. Generate a magnetic field. In order to make it possible to generate such a gradient magnetic field, the gradient coil unit 106 has three gradient coils (not shown). A gradient coil drive unit 130 is connected to the gradient coil unit 106, and the gradient coil drive unit 130 gives a drive signal to the gradient coil unit 106 to generate a gradient magnetic field. The gradient coil drive unit 130 has three drive circuits (not shown) corresponding to the three gradient coils in the gradient coil unit 106.

  The gradient magnetic field in the slice axis direction is called a slice gradient magnetic field, the gradient magnetic field in the phase axis direction is called a phase encode gradient magnetic field (or phase encode gradient magnetic field), and the gradient magnetic field in the frequency axis direction is read out gradient magnetic field (or frequency encode) Gradient magnetic field).

  When the coordinate axes orthogonal to each other in the static magnetic field space are the X axis, the Y axis, and the Z axis, any of the axes can be a slice axis. In the present embodiment, the slice axis is the body axis direction of the subject SB is the Z-axis direction, one of the remaining two axes is the phase axis, and the other is the frequency axis. Note that the slice axis, the phase axis, and the frequency axis can have an arbitrary inclination with respect to the X, Y, and Z axes while maintaining the orthogonality therebetween.

  The RF coil unit 108 forms a high-frequency magnetic field for exciting spins in the body of the subject SB in the static magnetic field space. Formation of a high-frequency magnetic field is called transmission of an RF excitation signal, and the RF excitation signal is called an RF pulse. An RF coil drive unit 140 is connected to the RF coil unit 108. The RF coil drive unit 140 provides a drive signal to the RF coil unit 108, and the RF coil unit 108 transmits an RF pulse based on the drive signal. An electromagnetic wave generated by the excited spin, that is, a nuclear magnetic resonance signal is received by the RF coil unit 108. A data collection unit 150 is connected to the RF coil unit 108. The data collection unit 150 collects echo signals (or MR reception signals) received by the RF coil unit 108 as digital data.

  Specifically, the RF coil unit 108 can use a multi receiver 80 (see FIG. 10) using a plurality of receiving coils. The RF coil unit 108 performs spatially uniform RF transmission / reception, and RF reception is performed by a multi-receiver (an RF receiving coil wound around the body) different from the RF coil unit. The multi-receiver 80 has a configuration in which a plurality of relatively highly sensitive receiving coils are arranged. The multi-receiver 80 synthesizes signals acquired by the coils. As a result, the magnetic resonance imaging apparatus 10 expands the field of view while maintaining the high sensitivity of the receiving coil to increase the sensitivity.

  The nuclear magnetic resonance signal detected by the RF coil unit 108 and collected by the data collecting unit 150 becomes a signal in the frequency domain (frequency domain), for example, Fourier space. The nuclear magnetic resonance signal is encoded in two axes by the gradient in the phase axis direction and the frequency axis direction. For this reason, the nuclear magnetic resonance signal can be obtained as a signal in a two-dimensional Fourier space, for example, when the frequency space is illustrated as a Fourier space. The two-dimensional Fourier space is also referred to as k-space. The phase encoding and magnetic field encoding (readout) gradient fields determine the sampling position of the signal in two-dimensional Fourier space.

  A sequence control unit 160 is connected to the gradient coil drive unit 130, the RF coil drive unit 140, and the data collection unit 150.

  The sequence control unit 160 drives the gradient coil driving unit 130 and the RF coil driving unit 140 in accordance with the imaging conditions input by the operator, that is, the imaging protocol.

  The display unit 180 is configured by a graphic display or the like. The display unit 180 is connected to the data processing unit 170. The display unit 180 can display an operation screen, an image reconstructed image, and the like.

  The operation unit 190 includes a keyboard having a pointing device. The operation unit 190 is connected to the data processing unit 170. The operation unit 190 is operated by the operator via the display unit 180. The operation unit 190 may arrange a touch panel on the display unit 180 instead of a keyboard or the like.

  The data processing unit 170 includes an arithmetic unit 171 and a storage unit 172. The data processing unit 170 executes various data processing and programs. The data processing unit 170 may be a workstation connected to a network different from the magnetic resonance imaging apparatus 10. The calculation unit 171 includes a function acquisition unit 175 and an image reconstruction unit 176. The function acquisition unit 175 acquires the correlation between the k-space reference data and the undersampling data, and the image reconstruction unit 176 creates an image by performing Fourier transform FT on the k-space imaging data. The storage unit 172 stores various shooting protocols, various programs, and various data, and appropriately reads and writes data and the like.

  Many nuclear magnetic resonance signals collected by the data collection unit 150 are input to the data processing unit 170. The data processing unit 170 causes the storage unit 172 to store the shooting data collected by the data collection unit 150. In the storage unit 172, a data space corresponding to the k space described above is formed. Hereinafter, the photographing data processing method in this embodiment will be described.

<Shooting data processing method>
In general, the magnetic resonance imaging apparatus 10 performs imaging with a plurality of types of pulse sequences on the same slice plane. The magnetic resonance imaging apparatus 10 acquires a plurality of contrast images by changing the parameters of the pulse sequence. For example, the magnetic resonance imaging apparatus 10 captures the same slice plane with a T1-weighted image, a T2-weighted image, a proton density image, and the like. An observer such as a doctor diagnoses a disease by comparing a plurality of images having different contrasts on the same slice plane.

  For example, the magnetic resonance imaging apparatus 10 captures a T1-weighted image, a T2-weighted image, and a proton density image on the same slice plane in a series of photographing of a predetermined part. The T1-weighted image, the T2-weighted image, and the proton density image are photographed using corresponding pulse sequences, and photographing data is collected. For this reason, in order to obtain a T1-weighted image, a T2-weighted image, and a proton density image, photographing is performed three times. Each two-dimensionally arrayed k-space imaging data is densely filled, and three times of full sampling imaging data is collected. Although the three full-sampled shooting data have different shooting times (collection times), about three times the shooting time is required for full sampling. For this reason, a series of imaging of a predetermined part requires three times of data amount and about three times the imaging time.

  In the proton density image, the density of hydrogen atoms is imaged. The proton density image has a low image contrast, but since the body of the subject contains a large amount of hydrogen atoms, the amount of imaging data is large and the signal to noise ratio (SNR) is excellent, which is useful for morphological information.

  The T1-weighted image is an image in which the T1 relaxation time of the proton density image is emphasized, and the T2-weighted image is an image in which the T2 relaxation time of the proton density image is emphasized. In other words, the T1-weighted image and the T2-weighted image are images in which the contrast of the proton density image is changed, and can be expressed using a conversion formula. Therefore, the T1-weighted image and the T2-weighted image are converted from the proton density image using the conversion formula. Can be created. The T1-weighted image can also be expressed as an image obtained by extracting the T1 relaxation time component of the proton density image, and the T2-weighted image can also be expressed as an image obtained by extracting the T2 relaxation time component of the proton density image.

  In the magnetic resonance imaging apparatus 10, an image to be reconstructed is created by performing Fourier transform FT on k-space imaging data. A high-quality image is created in a dense state (full sampling) in which the entire area of the two-dimensionally arranged k-space is filled with photographing data. On the other hand, compressed sensing is a technique for reconstructing an image even in a sparse state (under sampling) in the k-space. Compressed sensing reconstructs an image even in undersampling shooting data and speeds up shooting, but it is difficult to say that it is a sampling method in which image quality is hardly deteriorated because the amount of shooting data is also reduced.

  As an example of the sampling method, there is a method of obtaining by a sampling pattern using a probability density function in k-space. In the sampling pattern, a sampling area in the k space is calculated by a probability density function.

  FIG. 2A is an example of imaging data in k space at a predetermined part. As shown in the drawing, the photographing data indicated by kx which is the X axis and ky which is the Y axis is shown to be closer to white as the signal intensity is higher and closer to black as the signal intensity is lower. Note that high signal strength means that the amplitude of the echo signal is large. Accordingly, when observing FIG. 2A, it is understood that the signal intensity at the center of k-space and the periphery thereof is large, and the signal intensity decreases as the edge of the k-space is approached. A probability density function is calculated from the signal intensity distribution, and FIG. 2B shows an example of the probability density distribution on the kx line of the probability density function. Note that kx is the frequency encoding direction, and ky is the phase encoding direction.

  The sampling pattern specifies a region in the probability density function where the probability is greater than the threshold value TH. FIG. 3 is a diagram illustrating a region in the probability density function where the probability is greater than the threshold value TH.

  A region R in which the probability is equal to or higher than the threshold value TH is a sampling pattern, and imaging data is collected in the region R by a predetermined pulse sequence. The area below the threshold TH is an unsampled area that is not sampled. In the present embodiment, such sampling is referred to as undersampling, as compared with full sampling in which imaging data in k space is densely filled. In undersampling, a part of k-space shooting data is not sampled, and k-space shooting data is sparse.

  For example, undersampling is sampling in which only the k-space region of the region R shown in FIG. 3 is acquired and regions other than the region R are missing. Such undersampling reduces the scanning time. However, even if the compression sensing method is used, artifacts remain in the image reconstructed based on the undersampled k-space data. Therefore, a high-quality image is acquired using the method described below.

  In the present embodiment, the undersampled T1-weighted image and T2-weighted image are compared with the proton density image, and unsampled imaging data is obtained by calculation. In this embodiment, high-quality T1-weighted images and T2-weighted images are reconstructed.

  In the following, a method for creating a T1-weighted image from a proton density image will be described as a representative. Note that images created from proton density images can be created not only in undersampled T1-weighted images and T2-weighted images, but also in in-phase images and out-of-phase images. The in-phase image refers to an image of the sum of water and fat hydrogen atoms, and the out-of-phase image refers to an image of the difference between fat hydrogen atoms and water hydrogen atoms.

<Creation procedure>
As described above, the T1-weighted image can be regarded as an image in which the contrast of the proton density image is changed. In the present embodiment, the proton density image is described as a reference image RI (Ref Image), and the T1-weighted image is described as a contrast image CI (Contrast Image). In addition, the contrast image CI is the same as a T2-weighted image, an in-phase image, and an out-of-phase image.

FIG. 4 is a flowchart showing a procedure for acquiring the contrast image CI2.
In step S11, the operator acquires shooting data ri of the reference image RI. The data collection unit 150 collects imaging data using a pulse sequence for full sampling of the reference image RI. The imaging data ri of the reference image RI is stored in a two-dimensionally arranged k space in the memory 172.

  FIG. 5A is a schematic diagram of the k space of the reference image RI arranged two-dimensionally. As shown in the drawing, since the photographing data ri of the reference image RI is fully sampled, the photographing data is dense in the k space. Note that the solid line shown in the drawing indicates sampled image data of a region in the k space (kx and ky), where kx is a frequency encoding direction and ky is a phase encoding direction.

  FIG. 5B is a schematic diagram of the reconstructed reference image RI. The image reconstruction unit 176 forms the reference image RI shown in FIG. 5B by reconstructing the captured data ri in FIG. 5A. The image reconstruction of the reference image RI can be created by performing Fourier transform FT on the imaging data fully sampled by the image reconstruction unit 176 of the data processing unit 170. Since the reference image RI has a large amount of information as a basis for the reference image RI in the subject, it is possible to collect imaging data in a relatively short time. The reference image RI can acquire a high-quality image having an excellent SNR, but has a low contrast.

  Returning to FIG. 4, in step S12, the operator obtains the photographing data ci1 of the contrast image CI1. The data collection unit 150 collects imaging data using an undersampling pulse sequence of the undersampled contrast image CI1. Shooting data ci1 of the contrast image CI1 is stored in a two-dimensionally arranged k-space in the memory 172.

  FIG. 6A is a schematic diagram of the k space of the two-dimensionally arranged contrast image CI1. As shown in the drawing, since the photographing data ci1 of the contrast image CI1 is undersampled, the photographing data is sparse in the k space. As shown in the drawing, the solid line indicates the sampled image data of the region in the k space (kx and ky), and the broken line indicates an unsampled state in which the image data is not sampled. In order to help understanding, in FIG. 6A, a solid line and a broken line are drawn every other line in the phase encoding direction. In practice, it is preferable to undersample the region where the amplitude of the echo signal shown in FIG. As shown in FIG. 3, the region where the signal intensity is large is generally the central region of k-space.

  FIG. 6B is a schematic diagram of the contrast image CI1 reconstructed. The image reconstruction unit 176 of the data processing unit 170 forms the reference image CI1 shown in FIG. 5B by reconstructing the imaged data ci1 of FIG. 6A using the compression sensing technique. . Since the contrast image CI1 is based on undersampled photographing data, an image that is deteriorated in image quality and is not suitable for a doctor's observation may be formed.

  Returning to FIG. 4, in step S <b> 13, the calculation unit 171 calculates the correlation between the shooting data ci <b> 1 of the contrast image CI <b> 1 and the shooting data ri of the reference image RI.

The correlation of the photographing data is calculated by the function acquisition unit 175 of the calculation unit 171. The correlation is such that the image function of the reference image RI is ref (x, y), the image function of the contrast image CI1 is con (x, y), and the contrast conversion function is f (x, y). The relational expression of Formula 1 is established.
ref (x, y) = con (x, y) / f (x, y)...
Equation 1 is transformed into Equation 2.
con (x, y) = f (x, y) × ref (x, y)...

Expression 2 in the image space has a relationship shown in FIG.
Further, the imaging data ci1 of the undersampled contrast image CI1 in k space is CON (kx, ky), the imaging data ri of the fully sampled reference image RI is REF (kx, ky), and the correlation function is F ( kx, ky). Equation 2 can be expressed in the form of convolution as shown in Equation 3 in k-space.

CON (kx, ky) = F (kx, ky) * REF (kx, ky) ... Formula 3
However, * is convolution (convolution integration).

  The kernel (integral kernel) F (kx, ky) expressed by Equation 3 can be calculated from the shooting data ri of the reference image RI and the shooting data ci1 of the contrast image CI1.

  That is, the correlation function kernel F (kx, ky) is calculated by analyzing the undersampled imaging data ci1 of the contrast image CI1 and the fully sampled imaging data ri of the reference image RI.

  FIG. 8 is a schematic diagram of the imaging data ci1 of the undersampled contrast image CI1 in the k space, the imaging data ri of the fully sampled reference image RI, and the correlation function f1 (kx, ky). Since the point c1 (x, y) of the k-space shooting data of the contrast image CI1 has a correlation with the region r1 (x, y) of the shooting data ri of the fully sampled reference image RI, the correlation function f1 (kx , Ky) is obtained. Further, since the point c2 (x, y) of the imaging data ri in the k space of the different contrast image CI1 is correlated with the region r2 (x, y) of the imaging data ri of the fully sampled reference image RI, the correlation function f2 A simultaneous equation of (kx, ky) is obtained. Similarly, the point cm (x, y) of the imaging data ri in the k-space of the mth contrast image CI1 is obtained as fm (kx, ky) from the region rm (x, y) of the imaging data ri of the reference image RI. . m is an integer of 3 or more. In FIG. 8, only points c1 (x, y) and r1 (x, y) are shown.

  The function acquisition unit 175 increases the number of simultaneous equations from the unknown, so that a plurality of correlation functions f1 (kx, ky), f2 (kx, ky), ... fm (kx, ky), ... fn (kx , Ky), the simultaneous equations are solved. Then, the function acquisition unit 175 acquires the kernel F (kx, ky).

  Returning to FIG. 4, in step S14, the image reconstruction unit 176 calculates unsampled shooting data from the kernel F (kx, ky).

  FIG. 9A is a schematic diagram of the imaging data CON (kx, ky) of the undersampled contrast image CI1, the kernel F (kx, ky), and the imaging data REF (kx, ky) of the reference image RI. It is. Based on the kernel F (kx, ky) and the shooting data REF (kx, ky) of the reference image RI, the image reconstruction unit 176 calculates the unsampled shooting data d1 (x, y) of the contrast image CI1. Is possible. Unsampled shooting data d2 (x, y), d3 (x, y),... Dn (x, y) of other contrast images CI1 (not shown) are also converted by the image reconstruction unit 176 into the kernel F (kx, ky). ) And the imaging data REF (kx, ky) of the reference image RI.

  The image reconstruction unit 176 obtains all of the unsampled shooting data d1 (x, y), d2 (x, y),... Dn (x, y) of the contrast image CI1. Then, as shown in FIG. 9B, the shooting data Ci1 of the contrast image CI1 that was sparse in the k space is supplemented with the shooting data to become dense shooting data ci2 in the k space.

  Returning to FIG. 4, in step S15, the image reconstructing unit 176 reconstructs the contrast image CI2 from the dense photographing data ci2 of the contrast image CI1. FIG. 9C is a diagram showing a contrast image CI2 obtained by performing a Fourier transform FT using the dense photographing data ci2. Since the image reconstruction unit 176 can perform the Fourier transform FT using the dense photographing data ci2, it can reconstruct a high-quality contrast image CI2.

In step S16, it is determined whether another contrast image CI is acquired.
If another contrast image CI is to be acquired, the process returns to step S12. If another contrast image CI is not to be acquired, the process ends.

  Steps S13 to S15 can be performed at a workstation connected to the network without using the data processing unit 170 provided in the magnetic resonance imaging apparatus 10. The flowchart shown in FIG. 4 can be stored in a storage medium as a program. By installing the program stored in the storage medium by a computer, the data processing unit 170 or the workstation can be processed.

  The image reconstruction method described above can be used simultaneously with an imaging method using a parallel imaging method. The parallel imaging method is a high-speed imaging method using a difference in sensitivity of multi-coils, and includes a SENSE (Sensitivity Encoding) method and a GRAPPA (Generalized Auto calibrating Partially Parallel Acquisition) method. The SENSE method is an image reconstruction method used for an image after Fourier transform, and GRAPPA is an image reconstruction method used on the k space. For example, in the parallel imaging method, imaging steps are shortened by detecting the nuclear magnetic resonance signal from the subject by thinning out the encoding step in the phase direction, and further, the imaging field is narrowed and a false image (artifact) is generated. Generate an existing image. In the double parallel imaging method, the false image is removed based on the difference in sensitivity distribution of the number of coils, and an image with a wide field of view is finally obtained.

  Multi-receiver 80 is used for parallel imaging. FIG. 10A shows an example of the multi-receiver 80. The multi-receiver 80 is composed of eight 8-shaped receiving coils 81 to 88, and two of them are arranged facing each other on the XY plane and the YZ plane with a predetermined distance therebetween. The multi-receiver 80 receives a nuclear magnetic resonance signal from a subject placed in a space surrounded by these receiving coils.

  In order to receive a nuclear magnetic resonance signal, an optimal combination of two receiving coils is selected from the eight receiving coils based on the imaging section and the phase encoding direction. For example, in the arrangement of the multi-receiver 80 shown in FIG. 10A, consider a case where the YZ plane or the YX plane is photographed with the Y direction as the phase encoding direction. In this case, as shown in FIG. 10B, the first receiving coil group 81, 83, 85, 87 is one of the optimum combinations. As shown in FIG. 10C, the second receiving coil group 82, 84, 86, 88 is one of the optimum combinations.

  Specifically, the optimal receiving coil is that when the sensitivity distributions of the first receiving coil group and the entire second receiving coil group are combined, there is no low region of the sensitivity distribution in the phase encoding direction, and the first receiving coil group And the second receiving coil group are combined in consideration of the fact that the sensitivity distributions are not the same. Note that the optimal method of combining the receiving coils is determined by the imaging section and the phase encoding direction if the arrangement of the receiving coils is constant. When which slice plane is to be photographed is set, the optimum receiving coil combination is automatically selected.

  In the multi-receiver 80, the kernel F (kx, ky) does not change in each receiving coil. Therefore, the kernel F (kx, ky) used in the first receiving coil group 81 can be applied to the image reconstruction of the contrast image CI of the second receiving coil group 82. Therefore, the multi-receiver 80 can increase the simultaneous type by a multiple of the receiving coil.

DESCRIPTION OF SYMBOLS 10 Magnetic resonance imaging apparatus 80 Multi receiver 100 Magnet system 102 Main magnetic field coil part 106 Gradient coil part 108 Coil part 130 Gradient coil drive part 140 Coil drive part 150 Data collection part 160 Sequence control part 170 Data processing part 171 Calculation part 172 Storage part 175 Function acquisition unit 176 Image reconstruction unit 180 Display unit 190 Operation unit 200 Bed
CI Contrast image FT Fourier transform RI Reference image SB Subject TH Threshold

Claims (6)

  A magnetic resonance imaging apparatus that executes a first pulse sequence for sampling k-space data and a second pulse sequence for sampling under-sampling data lacking a part of k-space data,
  Based on the k-space data obtained in the first pulse sequence and the under-sampling data obtained in the second pulse sequence, the k-space data and the under-sampling data obtained in the first pulse sequence. Means for obtaining a correlation function with
  Based on the correlation function, reconstructing means that compensates for missing portion data in the undersampling data and reconstructs an image using k-space data after the data is supplemented;
With
  In the first pulse sequence, k-space data for obtaining a proton density image is sampled,
  The magnetic resonance imaging apparatus in which undersampling data for obtaining a T1-weighted image, a T2-weighted image, an in-phase image, or an out-of-phase image is sampled in the second pulse sequence.   The means for obtaining the correlation function is:
  The correlation function is obtained based on k-space data obtained by the first pulse sequence and data in a central region of k-space among the undersampling data obtained by the second pulse sequence. Item 2. The magnetic resonance imaging apparatus according to Item 1.   The means for obtaining the correlation function is:
  The correlation function is obtained based on k-space data obtained by the first pulse sequence and data of a region having a large amplitude in k-space among the undersampling data obtained by the second pulse sequence. The magnetic resonance imaging apparatus according to claim 1.   When k-space data obtained by the first pulse sequence is REF (kx, ky) and the undersampling data obtained by the second pulse sequence is CON (kx, ky), The magnetic resonance imaging apparatus according to claim 1, wherein a satisfying kernel (integral kernel) F (kx, ky) is the correlation function.
CON (kx, ky) = F (kx, ky) * REF (kx, ky)...
However, * is convolution (convolution integration).   5. The pulse sequence according to claim 1, wherein at least one of the first pulse sequence and the second pulse sequence is a pulse sequence using a parallel imaging method. 6. Magnetic resonance imaging device.   A first pulse sequence for sampling k-space data, the first pulse sequence for sampling k-space data for obtaining a proton density image, and an under lacking a part of k-space data A second pulse sequence for sampling sampling data, the second pulse sequence for sampling undersampling data for obtaining a T1-weighted image, a T2-weighted image, an in-phase image, or an out-of-phase image; A program applied to a magnetic resonance imaging apparatus to be executed,
  Based on the k-space data obtained in the first pulse sequence and the under-sampling data obtained in the second pulse sequence, the k-space data and the under-sampling data obtained in the first pulse sequence. A process for obtaining a correlation function with
  Based on the correlation function, reconstructing processing that compensates for missing data in the undersampling data and reconstructs an image using k-space data after the data is supplemented;
A program that causes a computer to execute.