JP6013882B2 - Magnetic resonance imaging apparatus and water fat separation method - Google Patents

Description

  The present invention relates to a nuclear magnetic resonance imaging (hereinafter referred to as MRI) technique, and more particularly to a water / fat separation imaging technique.

  MRI images the form or function of the human head, abdomen, limbs, etc. in two or three dimensions by imaging the spatial distribution of relaxation information of proton density and nuclear magnetic resonance signals. In the current MRI, the radionuclide to be imaged that is widely used clinically is a hydrogen nucleus (proton) that is a main constituent of the subject.

  When a human body is a measurement target, water and fat are main protons that can be detected by MRI. However, fat signals present in the human body reduce the image contrast in the subject's abdomen, spine, limbs, and the like. Therefore, methods for suppressing fat signals have been proposed in the clinic.

  As one of the methods, a Dixon method for suppressing fat using a phase difference between water and fat is known (for example, see Non-Patent Document 1). Water and fat have “chemical shifts” that differ in resonance frequency by −3.5 ppm due to differences in molecular structure. The frequency difference due to this chemical shift is proportional to the magnetic field strength, and is approximately -224 Hz when the magnetic field strength is 1.5 Tesla. The Dixon method separates a water signal and a fat signal in an acquired image using a phase difference between water and fat nuclear magnetization vectors caused by a difference in frequency between water and fat. Thereby, the image (water image) which consists of water signals in which fat is suppressed is obtained. An image composed of the fat signals after separation is called a fat image.

In the Dixon method, two pieces of image data having different time (hereinafter referred to as echo time) from excitation of a nuclear spin to acquisition of a signal are acquired. These two image data are an original image I _in acquired at an echo time t _in when water and fat are in phase, and an echo time t _out at which water and fat are out of phase. The acquired original image _Iout . By adding and subtracting these, the signal intensity W of the water image and the signal intensity F of the fat image are calculated, respectively, and the water signal and the fat signal are separated.

  However, the Dixon method does not take into account the phase change caused by the static magnetic field inhomogeneity that occurs when the subject is inserted into the static magnetic field space. When spatial static magnetic field inhomogeneity exists, a phase change different from the chemical shift occurs depending on the position. Therefore, water and fat cannot be completely separated by simple addition and subtraction processing.

  Therefore, a method called a 3-Point Dixon method is known as a water / fat separation method in consideration of inhomogeneous static magnetic fields (see, for example, Non-Patent Document 2). The 3-Point Dixon method uses a least-square estimation process from three images with different echo times for three variables: a signal intensity W of a water image, a signal intensity F of a fat image, and a frequency difference due to static magnetic field nonuniformity. Is a method of separating water and fat by determining three variables by repeating the above. However, since the 3-Point Dixon method needs to perform three measurements, the measurement time is prolonged.

As a technique that takes static magnetic field inhomogeneity into account and does not prolong measurement time, a technique called a 2-Point Dixon method is known (for example, see Non-Patent Document 3). In the 2-Point Dixon method, similarly to the Dixon method, an echo time t _in where water and fat are in phase and an echo time t _out where water and fat are out of phase. The original image I _in and the original image I _out are obtained, respectively. Thereafter, a phase change (hereinafter simply referred to as a phase distribution) caused by the static magnetic field inhomogeneity is estimated by performing phase unwrapping using a region enlargement method. Then, by adding and subtracting the original images I _in and I _out after removing the phase change, the water image and the fat image are separated.

In the 2-Point Dixon method, after measuring the original image I _in at the echo time t _in when water and fat are in phase, the gradient magnetic field pulse is inverted, and the water and fat are _out of phase at the echo time t _out. The original image _Iout is acquired. That is, an image having the same phase and the opposite phase is acquired by one measurement.

Thomas Dixon, et al. "Simple Proton Spectroscopic Imaging" Radiology, 153, 1984, 189-194 Scott B. Reeder, et al. "Multicoil Dixon Chemical Specialization With an Iterative Least-Squares Estimation Method" Magnetic Resonance in Medicine, 51, 35-45. Jingfei Ma “Breath-Hold Water and Fat Imaging Using a Dual-Echo Two-Point Dixon Technological and Robust Phase-Corporation”

  The 2-Point Dixon method described in Non-Patent Document 3 obtains an image in which water and fat are in the same phase and an image in opposite phase by reversing the gradient magnetic field pulse, so that the measurement time is It will not last long. However, when calculating the phase distribution, if the starting point of the unwrapping process (hereinafter referred to as a seed point) is set to a pixel whose main component is fat, the obtained phase distribution is generated by the chemical shift of fat. The offset of the phase value π is added. If a water image and a fat image are calculated using this phase distribution, they are interchanged. Therefore, according to the 2-Point Dixon method described in Non-Patent Document 3, the user must determine which of the separated images is a water image and which is a fat image each time an image is taken.

  The present invention has been made in view of the above circumstances, and in the 2-Point Dixon method, the water image and the fat image are interchanged by improving the accuracy of correctly calculating the phase image in which the static magnetic field inhomogeneity is reflected. To provide a technique for separating water and fat without burdening the operator with judgment.

  The present invention increases the possibility of correctly calculating the phase distribution, which is a phase change caused by static magnetic field inhomogeneity. Then, water and fat are separated using a signal value from which the influence of the phase distribution due to the static magnetic field inhomogeneity is removed. In order to realize this, in the calculation of the phase distribution, start point candidates are specified so that the probability that a pixel mainly composed of a water signal is selected as the start point of the phase unwrapping process using the region enlargement method is increased. Furthermore, the correctness of the calculated phase distribution is determined with high accuracy using the characteristics of the starting point candidate.

  That is, a magnetostatic generating magnet that generates a static magnetic field in a space in which the subject is placed, a transmission unit that transmits a high frequency magnetic field pulse to the subject, and a nuclear magnetic resonance signal generated from the subject by transmission of the high frequency magnetic field pulse Controls the operations of the receiving unit for receiving, the gradient magnetic field applying unit for applying a gradient magnetic field for adding position information to the nuclear magnetic resonance signal, the transmission unit, the gradient magnetic field applying unit, and the receiving unit. A computer for performing arithmetic processing on the received nuclear magnetic resonance signal, the computer obtaining the nuclear magnetic resonance signal of two different echo times according to a predetermined pulse sequence A measurement unit, a reconstruction unit for reconstructing the original images of the two different echo times from the two nuclear magnetic resonance signals, and an original image of the two different echo times Water that separates a water signal and a fat signal by using a phase distribution calculation unit that calculates a phase distribution that is a phase distribution that changes due to static magnetic field inhomogeneity, the original image of the two different echo times, and the phase distribution. A fat separation unit, wherein the phase distribution calculating means uses the original images of the two different echo times and specifies a pixel group affected by noise as a seed point candidate pixel group; , Using the original images of the two different echo times and the information of the seed point candidate pixel group, a temporary phase distribution calculating unit that calculates a temporary phase distribution that is the temporary phase distribution, and whether the temporary phase distribution is correct or incorrect There is provided a magnetic resonance imaging apparatus comprising: a correct / incorrect determination unit for determining; and a true phase distribution determination unit for determining a true phase distribution based on a determination result by the correct / incorrect determination unit.

  Also, water / fat separation for separating a water signal and a fat signal from the original images of the two different echo times respectively reconstructed from nuclear magnetic resonance signals of two different echo times measured according to a predetermined pulse sequence The method uses the original images of the two different echo times, identifies a group of pixels affected by noise as a seed point candidate pixel group, and the original images of the two different echo times and the seed point candidate pixels Using the group information, calculating the provisional phase distribution that is the provisional phase distribution, determining the correctness of the provisional phase distribution, determining the true phase distribution based on the determination result by the correctness determination unit, A water / fat separation method is provided that separates a water signal and a fat signal using an original image of two different echo times and the true phase distribution.

  According to the present invention, it is possible to prevent the water image and the fat image from being switched by increasing the accuracy of correctly calculating the phase image in which the static magnetic field inhomogeneity is reflected, and without burdening the operator to make a determination, Can separate water and fat.

FIG. 3A is an external view of a horizontal magnetic field type MRI apparatus in the MRI apparatus according to the embodiment of the present invention. FIG. 2B is an external view of a vertical magnetic field type MRI apparatus in the MRI apparatus according to the embodiment of the present invention. (C) is an external view of the MRI apparatus in which the tunnel type magnet is inclined obliquely in the MRI apparatus of the embodiment of the present invention. It is a functional block diagram of the MRI apparatus of embodiment of this invention. It is a functional block diagram of the computer of the embodiment of the present invention. It is a flowchart of the measurement process of embodiment of this invention. It is explanatory drawing for demonstrating an example of the gradient echo type | mold pulse sequence of embodiment of this invention. It is explanatory drawing for demonstrating an example of the spin echo type | mold pulse sequence of embodiment of this invention. It is a flowchart of the phase distribution calculation process of embodiment of this invention. It is a functional block diagram of the phase distribution calculation part of this invention. It is a flowchart of the seed point candidate pixel group specifying process of the embodiment of the present invention. It is a graph of signal ratio distribution of the embodiment of the present invention. It is a flowchart of the temporary phase distribution calculation process of embodiment of this invention. It is a flowchart of the true phase distribution determination process of the embodiment of the present invention.

  Hereinafter, embodiments to which the present invention is applied will be described. Hereinafter, in all the drawings for explaining the embodiments, the same reference numerals are given to components having the same functions unless otherwise mentioned, and the repeated explanation thereof is omitted.

  First, the magnetic resonance imaging apparatus (MRI apparatus) of this embodiment will be described. FIG. 1 is an external view of the MRI apparatus of this embodiment. FIG. 1A shows a horizontal magnetic field type MRI apparatus 100 using a tunnel magnet that generates a static magnetic field with a solenoid coil. FIG. 1B shows a hamburger type (open type) vertical magnetic field type MRI apparatus 120 in which magnets are separated into upper and lower sides in order to enhance the feeling of opening. FIG. 1C shows an MRI apparatus 130 that uses the same tunnel-type magnet as that in FIG. 1A, shortens the depth of the magnet, and tilts it obliquely to enhance the feeling of opening. In the present embodiment, any of these MRI apparatuses having these appearances can be used. These are merely examples, and the MRI apparatus of the present embodiment is not limited to these forms. In the present embodiment, various known MRI apparatuses can be used regardless of the form and type of the apparatus. Hereinafter, when there is no need to distinguish between them, the MRI apparatus 100 is representative.

  FIG. 2 is a functional configuration diagram of the MRI apparatus 100 of the present embodiment. As shown in the figure, the MRI apparatus 100 of the present embodiment generates a static magnetic field in a space where the subject 101 is placed, for example, a static magnetic field generating magnet such as a static magnetic field coil (hereinafter referred to as a static magnetic field coil). 102, a shim coil 104 that adjusts the static magnetic field distribution, a transmission high-frequency coil 105 that transmits a high-frequency magnetic field to the measurement region of the subject 101 (hereinafter simply referred to as a transmission coil), and a nuclear magnetic resonance signal generated from the subject 101 In order to add position information to the receiving magnetic coil 106 (hereinafter simply referred to as a receiving coil) and a nuclear magnetic resonance signal generated from the subject 101, gradient magnetic fields are applied in the x, y, and z directions, respectively. Gradient magnetic field coil 103, transmitter 107, receiver 108, calculator 109, gradient magnetic field power supply 112, shim power supply 113, and sequence And a control unit 114, a.

  Various types of static magnetic field coils 102 are adopted according to the structures of the MRI apparatuses 100, 120, and 130 shown in FIGS. 1 (a), 1 (b), and 1 (c), respectively. . The gradient magnetic field coil 103 and the shim coil 104 are driven by a gradient magnetic field power supply unit 112 and a shim power supply unit 113, respectively. In this embodiment, a case where separate transmission coils 105 and reception coils 106 are used will be described as an example. However, the transmission coil 105 and the reception coil 106 are configured as a single coil. May be. The high-frequency magnetic field irradiated by the transmission coil 105 is generated by the transmitter 107. The nuclear magnetic resonance signal detected by the receiving coil 106 is sent to the computer 109 through the receiver 108.

  The sequence controller 114 controls the operations of the gradient magnetic field power supply unit 112 that is a drive power supply for the gradient coil 103, the shim power supply unit 113 that is the drive power supply for the shim coil 104, the transmitter 107, and the receiver 108. Controls the application of a gradient magnetic field, a high-frequency magnetic field, and reception of a nuclear magnetic resonance signal. The control time chart is called a pulse sequence, is preset according to measurement, and is stored in a storage device or the like included in the computer 109 described later.

  The computer 109 controls the overall operation of the MRI apparatus 100 and performs various arithmetic processes on the received nuclear magnetic resonance signal. In this embodiment, an original image having a different echo time, a separated water image, a fat image, and a phase distribution indicating a phase change due to static magnetic field inhomogeneity are generated. The computer 109 is an information processing apparatus including a CPU, a memory, a storage device, and the like, and a display 110, an external storage device 111, an input device 115, and the like are connected to the computer 109.

  The display 110 is an interface for displaying results obtained by the arithmetic processing to the operator. The input device 115 is an interface for an operator to input conditions, parameters, and the like necessary for the arithmetic processing performed in the present embodiment. The external storage device 111 holds, together with the storage device, data used for various arithmetic processes executed by the computer 109, data obtained by the arithmetic processes, input conditions, parameters, and the like.

  Hereinafter, functions realized by the computer 109 according to the present embodiment will be described. FIG. 3 is a functional block diagram of the computer 109 of this embodiment. The computer 109 of this embodiment includes a measurement unit 310, a reconstruction unit 320, a phase distribution calculation unit 330, and a water-fat separation unit 340.

The measurement unit 310 operates the sequence control device 114 in accordance with the pulse sequence and controls each part of the MRI apparatus 100 to perform measurement to obtain a nuclear magnetic resonance signal (echo signal) having a predetermined echo time. In the present embodiment, echo signals having two different echo times are obtained. As two different echo times, an echo time t _in where water and fat are in phase and an echo time t _out where water and fat are in opposite phases are used.

The reconstruction unit 320 arranges the acquired echo signal in the k space and reconstructs an image by Fourier transform. In the present embodiment, original images I _in and I _out at two different echo times t _in and t _out are reconstructed, respectively.

  The phase distribution calculation unit 330 calculates a phase distribution (phase distribution) that changes due to static magnetic field inhomogeneity from the calculated original image. In the present embodiment, the phase distribution is calculated using a ratio image obtained by complex division of two different echo time original images.

  The water-fat separation unit 340 separates the water signal and the fat signal from the original images having two echo times different from the phase distribution calculated by the phase distribution calculation unit 330. Then, a water image composed of a water signal and a fat image composed of a fat signal are generated.

  Various functions realized by the computer 109 are realized by a CPU loading a program held in the storage device into the memory and executing the program. In addition, at least one of the various functions realized by the computer 109 is an information processing apparatus independent of the MRI apparatus 100 and is realized by an information processing apparatus capable of transmitting and receiving data to and from the MRI apparatus 100. It may be.

  Prior to the description of the water and fat separation process of this embodiment, the principle of the water and fat separation process by the Dixon method and the outline of the process by the 2 point Dixon method will be described.

As described above, in the Dixon method, echo and the original image I _in acquired at time t _in, by adding and subtracting the original image I _out acquired at echo time t _out, the signal strength W and fat water image The signal intensity F of the image is calculated, and the water signal and the fat signal are separated.

Here, when the frequency difference between water and fat is df, the time t _in when water and fat are in phase is n / df, and the time t _out when water and fat are in opposite phase is (2n + 1) / ( 2df). Note that n is an integer. Therefore, when the difference between the echo times t _in and t _out is dt, dt = 1 / (2df).

Ｄｉｘｏｎ法では、この２つの信号を、以下の式（２−１）および式（２−２）のように加算および減算することによって、水画像の信号強度Ｗと脂肪画像の信号強度Ｆとをそれぞれ算出し、水信号と脂肪信号とを分離する。

When the signal intensity of the water image is W and the signal intensity of the fat image is F, when there is no phase change due to static magnetic field inhomogeneity, I _in and I _out are respectively expressed by the following equations (1-1) and (1 -2).

In the Dixon method, the signal intensity W of the water image and the signal intensity F of the fat image are obtained by adding and subtracting these two signals as in the following expressions (2-1) and (2-2). Each is calculated and a water signal and a fat signal are separated.

On the other hand, in the 2-Point Dixon method, as described above, a phase change caused by non-uniform static magnetic field (hereinafter simply referred to as phase distribution) is further subjected to phase unwrapping using a region enlargement method. Estimated by In phase unwrapping using the region enlargement method, first, a reference pixel is set as the start point (seed point), and the phase of the pixel adjacent to the seed point is changed according to the phase difference from the seed point. Is determined to be smooth. The pixel for which the phase has been determined is used as the next seed point, and the same processing is repeated until the phase of all the pixels in the image is determined, and the phase distribution is calculated. Then, the phase value of the resulting phase distribution, after removing from the original image I _in, or I _out, adds the original image I _in and I _out, by subtracting separated into a water image and a fat image.

  Hereinafter, the flow of the entire measurement process (measurement process) by each function of the computer 109 of the present embodiment will be briefly described. FIG. 4 is a flowchart for explaining the flow of the measurement processing of the present embodiment.

In the present embodiment, taking the original image I _in and I _out of the two different echo times predetermined t _in and t _out, to calculate the phase distribution due to the static magnetic field inhomogeneity from the original image I _in and I _out, the phase From the distribution and the original images _Iin and _Iout , a water image and a fat image obtained by separating water and fat are calculated. In this embodiment, when calculating the phase distribution due to non-uniform static magnetic fields, the possibility that a pixel whose main component is water is set as a seed point is increased. Also, the correctness of the obtained phase distribution is also determined.

First, the measurement unit 310 controls the sequence controller 114 according to a predetermined pulse sequence, and measures echo signals of two different echo times t _in and t _out (step S1001).

Then, reconstruction unit 320, the resulting echo signals, respectively, by Fourier transform arranged on the k-space, two different original image I _in and I _out of the echo time t _in and t _out respectively Reconfiguration is performed (step S1002).

The phase distribution calculation unit 330 calculates the ratio of the two original images I _in and I _out and calculates the distribution (phase distribution) of the phase change caused by the static magnetic field inhomogeneity from the phase information of the ratio image (step S1003). .

Thereafter, the water-fat separation unit 340 separates the water signal and the fat signal from the calculated phase distribution and the two original images I _in and I _out (step S1004), and the water image including the separated water signal. And a fat image made up of fat signals are displayed on the display 110 (step S1005). Note that the process of step S1004 is referred to as a water-fat separation process. The separated water image and fat image are called separated images.

  When the water image and the fat image are displayed on the display 110, information that makes the two distinguishable may be displayed together. The distinguishable information is, for example, captions such as “water image” and “fat image”. Further, if necessary, the phase distribution calculated in step S1003 may be displayed on the display 110.

  Here, an example of a pulse sequence executed by the measurement unit 310 in the measurement in step S1001 will be described. FIG. 5 shows an example of a time chart of the pulse sequence 510 employed in the present embodiment. This pulse sequence 510 is a gradient echo type pulse sequence. In the pulse sequence 510, RF is an RF pulse, Gs is a slice selection gradient magnetic field, Gp is a phase encoding gradient magnetic field, Gr is a readout gradient magnetic field, and an echo is an echo signal acquisition timing. Each is shown. The same applies to each pulse sequence in the present specification.

In the pulse sequence 510, the echo signal is measured in the following procedure within one repetition time. First, the RF pulse 511 is irradiated to excite the hydrogen nuclear spin of the subject 101. At this time, a slice selection gradient magnetic field (Gs) 512 is applied simultaneously with the RF pulse 511 in order to select a specific slice of the subject 101. Subsequently, a phase encoding gradient magnetic field (Gp) 513 for phase encoding is applied to the echo signal. Thereafter, a read gradient magnetic field (Gr) 514 is applied after time TE ₁ from the first RF pulse 511 irradiation, and an echo signal (first echo) 515 is measured. Further, after a time dt from the measurement of the first echo 515, a read gradient magnetic field (Gr) 516 having a reversed polarity is applied to measure an echo signal (second echo) 517. The echo time TE ₂ for measuring the second echo 517 satisfies TE ₂ = TE ₁ + dt.

As described above, the echo time TE ₁ when measuring the first echo 515 and the echo time TE ₂ when measuring the second echo 517 are the time t _in when the water and fat are in phase, and The time t _out is set so that water and fat have opposite phases. Here, when the frequency difference between water and fat is df, the time t _in when water and fat are in phase is n / df, and the time t _out when water and fat are in opposite phase is (2n + 1) / ( 2df). Note that n is an integer.

In the pulse sequence 510 of this embodiment, an echo time TE ₁ and an echo time TE ₂ that satisfy the above-described conditions are selected. Note that the measurement order of the echo times t _in and t _out does not matter.

  In the measurement of this embodiment, the pulse sequence 510 is repeated a predetermined number of times while changing the intensity of the phase encoding gradient magnetic field 513. For example, the number of echo signals necessary for image reconstruction is repeatedly acquired 128 times, 256 times, and the like. One original image (first original image) is formed by the first echo for the number of repetitions, and another one original image (second original image) is formed by the second echo for the number of repetitions. These are used as original images for calculation for calculating a water image and a fat image described later.

  FIG. 6 shows a time chart of another example of a pulse sequence employed in the present embodiment. This pulse sequence 520 is a spin echo type pulse sequence.

In the pulse sequence 520, the echo signal is measured in the following procedure within one repetition time. First, the RF pulse 521 is irradiated to excite the spin of the subject 101. At this time, a slice selection gradient magnetic field (Gs) 522 is applied simultaneously with the RF pulse 521 in order to select a specific slice of the subject 101. Subsequently, a phase encoding gradient magnetic field (Gp) 523 for phase encoding is applied to the echo signal, and an RF pulse 528 for inverting the spin is applied together with a slice selection gradient magnetic field (Gs) 529. After that, after the time TE ₁ from the first RF pulse 521 irradiation, the readout gradient magnetic field (Gr) 524 is applied to measure the echo signal (first echo) 525. Further, after a time dt from the measurement of the first echo 525, the readout gradient magnetic field (Gr) 526 with reversed polarity is applied to measure the echo signal (second echo) 527.

The echo time TE ₁ when measuring the first echo 525 and the echo time TE ₂ when measuring the second echo 527 are selected so as to satisfy the same conditions as in the gradient pulse sequence.

  In the example shown in FIGS. 5 and 6, in the case of the gradient echo type pulse sequence 510 and the case of the spin echo type pulse sequence 520, two signals having different echo times within one repetition time (here, Then, the first echo and the second echo) are measured. However, one echo signal may be measured in one repetition time. In this case, the first echo and the second echo are measured using two repetition times.

In step S1002, the reconstruction unit 320 arranges the echo signals of the echo times t _in and t _out measured in step S1001 on the k space, and performs Fourier transform. Accordingly, the reconstruction unit 320 calculates the original images I _in (first original image) and I _out (second original image) of two different echo times (t _in and t _out ), respectively.

Next, calculation of the phase distribution by the phase distribution calculation unit 330 in step S1003 will be described. First, the phase distribution calculation unit 330 of the present embodiment starts with a pixel (hereinafter referred to as a seed point) that starts calculation of a phase distribution in a ratio image obtained by complex division of the original images I _in and I _out of two different echo times. ). Then, using the phase value of the ratio image, the phase value of each pixel is determined from the seed point by the enlarged region method, and the phase distribution is calculated.

  In order to realize this, as shown in FIG. 3, the phase distribution calculation unit 330 of the present embodiment includes a seed point candidate specifying unit 331 that specifies a pixel group that is a seed point candidate, and a pixel group that is specified as a seed point candidate. A temporary phase distribution calculation unit 332 that calculates a temporary phase distribution using one pixel as a seed point, a correct / incorrect determination unit 333 that determines whether the temporary phase distribution is calculated by the temporary phase distribution calculation unit 332, A true phase distribution determining unit 334 that determines a true phase distribution based on the correctness determination result of the determining unit 333.

  An outline of the flow of the phase distribution calculation process of the present embodiment according to the above functions will be described. FIG. 7 is a flowchart for explaining the flow of the phase distribution calculation process of the present embodiment.

The seed point candidate specifying unit 331 uses the two original images I _in and I _out to specify a pixel group (seed point candidate pixel group) that is a seed point candidate (step S1101). Next, the temporary phase distribution calculation unit 332 performs phase unwrap processing using an enlarged region method using any one point in the identified seed point candidate pixel group as a seed point, and calculates a temporary phase distribution ( Step S1102). Next, the correctness determination unit 333 determines whether the temporary phase distribution is correct (step S1103). Then, the true phase distribution determining unit 334 determines the true phase distribution based on the correctness / incorrectness determination result (step S1104). Details of each process will be described below.

  First, the seed point candidate specifying process by the seed point candidate specifying unit 331 in step S1101 will be described. The seed point candidate specifying unit 331 specifies a seed point candidate pixel group using original images of two echo times. At this time, when selecting an arbitrary pixel as a seed point, the seed point candidate pixel group is determined so that the probability that a pixel mainly composed of water is selected is increased. In the present embodiment, pixels that are strongly influenced by noise are specified as a seed point candidate pixel group with a high probability that water is a main component pixel.

  In order to realize this, the seed point candidate specifying unit 331 of the present embodiment calculates a ratio image from original images of two different echo times and calculates an absolute value thereof, as shown in FIG. A calculation unit 411, a threshold setting unit 412 that sets a threshold for specifying a pixel group of a seed point candidate from the absolute value ratio image, and a pixel group that specifies a seed point candidate pixel group in the absolute value ratio image using the threshold A specifying unit 413.

  An outline of the flow of seed point candidate pixel group specifying processing by each part of the seed point candidate specifying unit 331 will be described. FIG. 9 is a flowchart for explaining the flow of the seed point candidate pixel group specifying process of the present embodiment.

First, the absolute value ratio image calculating section 411, an original image I _out obtained from the echo signals measured in the echo time t _out, and the complex division in the echo time t _in the original image obtained from the echo signal measured by I _in, A ratio image I is calculated (step S1201). Then, the absolute value is taken and an absolute value ratio image | I | is calculated (step S1202). The absolute value ratio image | I | is calculated by the following equation (3).
| I | = | _Iout / _Iin | (3)

  Next, the threshold value setting unit 412 sets at least one threshold value Th (step S1203). The pixel group specifying unit 413 specifies, as a seed point candidate pixel group, a pixel group having a pixel value greater than or equal to the threshold Th among the pixel values of the absolute value ratio image | I | (step S1204).

  Here, a method for setting the threshold Th in step S1203 by the threshold setting unit 412 will be described. The threshold Th is set so that a pixel group having a high ratio of water-main components in the original image or the ratio image is specified as the seed point candidate pixel group.

In the original image _Iin , the signal value of the pixel (water pixel) with 100% water signal (simply called the water signal) is S _win , and the signal value of the pixel (fat pixel) with 100% fat signal is S _fin ), the water signal in the original image I _out is S _wout , the fat signal is S _fout , and the standard deviation of the mode 0 noise according to the Gaussian distribution is σ. At this time, the signal ratio (water signal ratio) when the water signal S _wout is complex-divided by S _win is S _mw , and the signal ratio (fat signal ratio) when the fat signal S _fout is complex-divided by S _fin is S _mf. Then, the water signal ratio S _mw and the fat signal ratio S _mf are expressed by the following equations (4-1) and (4-2), respectively.
S _mw = S _wout / S _win (4-1)
S _mf = S _fout / S _fin (4-2)
The water signal ratio S _mw and the fat signal ratio S _mf correspond to the signal values of the water pixel and the fat pixel in the ratio image, respectively.

At this time, _{assuming that} the standard deviations of the water signal ratio S _mw and the fat signal ratio S _mf are σ _mw and σ _mf , respectively, according to the law of error propagation, σ _mw and σ _mf are respectively expressed by the following equations (5-1), It is represented by Formula (5-2).

なお、ＳＮＲ_wおよびＳＮＲ_fは、水信号Ｓ_wと脂肪信号Ｓ_fの信号対雑音比（ＳＮＲ：Ｓｉｇｎａｌ ｔｏ Ｎｏｉｓｅ Ｒａｔｉｏ）をそれぞれ表し、以下の式（７−１）および式（７−２）で定義される。
ＳＮＲ_w＝Ｓ_w／σ・・・（７−１）
ＳＮＲ_f＝Ｓ_f／σ・・・（７−２） Assuming that the signal intensity when measured at the echo times t _in and t _out does not change, assuming that S _win = S _wout = S _w and S _fin = S _fout = S _f , the equation (5-1) and the equation (5-2) is represented by the following formulas (6-1) and (6-2).

SNR _w and SNR _f represent the signal to noise ratio (SNR) of the water signal S _w and the fat signal S _f , respectively, and the following equations (7-1) and (7-2) Defined by
SNR _w = S _w / σ (7-1)
SNR _f = S _f / σ (7-2)

Usually, the original images I _in and I _{out in} the 2-Point Dixon method are obtained as a T ₁ weighted image using a gradient echo method or a spin echo method, or a T ₂ weighted image using a fast spin echo method. At this time, it is known that the fat signal S _f is always larger than the water signal S _w . Accordingly, the SNR _f of the fat signal S _f is higher than the SNR _w of the water signal S _w . That is, the SNR _f fat signal S _f, and the SNR _w water signal S _w, a relationship of the following equation (8).
SNR _w <SNR _f (8)
Therefore, from the equations (6-1) and (6-2), the standard deviation σ _mw of the water signal ratio S _mw is larger than the standard deviation σ _mf of the fat signal ratio S _mf . That is, the standard deviation σ _mw and the standard deviation σ _mf have the relationship of the following formula (9).
σ _mw > σ _mf (9)

For example, when the SNR _w of the water signal S _w is 10 and the SNR _f of the fat signal S _f is 20 which is twice the SNR _w , the fat signal ratio S _{mf is obtained} from the equations (6-1) and (6-2). The relationship between the standard deviation σ _mf of the water signal and the standard deviation σ _mw of the water signal ratio S _wf is expressed by the following equation (10).
σ _mf = σ _mw / 2 (10)

FIG. 10 shows a histogram (signal ratio distribution) of the number of pixels (number of voxels) corresponding to the signal ratio (pixel value of the ratio image I) of the water signal ratio S _mw and the fat signal ratio S _mf at this time. The horizontal axis in FIG. 10 represents the signal ratio, and the vertical axis represents the number of pixels (number of voxels). _{Reference numeral} 601 represents the distribution of the water signal ratio S _mw , and 602 represents the distribution of the fat signal ratio S _mf . Here, the number of pixels the water signal S _w and a fat signal S _f of a 10 ^4, respectively. Further, the mode value of the water signal ratio S _mw and the fat signal ratio S _mf at this time is 1.

As shown in FIG. 10, when the SNR _w is less than SNR _f, since the standard deviation sigma _mw water signal ratio S _mw is larger than the standard deviation sigma _mf fat signal ratio S _mf, a histogram of the water signal ratio S _{mw Has} a wider distribution than the histogram of the fat signal ratio S _mf . Therefore, it can be seen that the number of pixels of the water signal ratio S _mw is larger than the number of pixels of the fat signal ratio S _{mf in} the signal ratio of the _bottom part on both sides of the distribution. The base portion of the signal ratio distribution is a pixel group that is strongly affected by noise.

In the present embodiment, the threshold Th for specifying a pixel group having a high ratio of pixels whose main component is water is set using the characteristics of the distribution of the water signal ratio S _mw and the fat signal ratio S _mf . That is, the threshold value Th is set using the pixel value distribution of the absolute value ratio image. For example, as illustrated in FIG. 10, when the threshold Th603 is set to a signal ratio of 1.2, a pixel with a signal ratio of 1.2 or higher has a high probability of being a water pixel.

Here, it is assumed that 100% of water or fat exists in the pixel. However, in an actual living tissue, since water and fat are mixed at different ratios depending on the tissue, the assumption that the signal intensity when measured at the echo times t _in and t _out does not change often does not hold. Usually, S _win ≧ S _wout and S _fin ≧ S _fout , and the mode values of the water signal ratio S _mw and the fat signal ratio S _mf are smaller than 1. Therefore, the threshold Th is set in consideration of this. For example, the threshold value Th is set to 1.0 or more. Even when water and fat are mixed in different proportions depending on the tissue, the relationship between the number of pixels of the water signal ratio S _{mw and} the number of pixels of the fat signal ratio S _mf is the same as described above.

For example, the threshold value Th may be determined in advance and held in the external storage device 111 or the storage device of the computer 109. Further, for each measurement, the threshold setting unit 412 calculates the mode value and the standard deviation of the signal ratio distribution (histogram) for the absolute value ratio image | I |, and the sum of the mode value and the standard deviation, or Any value greater than that may be set as the threshold Th. An arbitrary value larger than the mode value may be set as the threshold Th. Further, the SNR _f of SNR _w and a fat signal S _f of the water signal S _w, the standard deviation of the distribution of the signal ratio is changed. Thus, the threshold setting unit 412 of the present embodiment, based on the SNR of the original image I _in, or I _out, Computer simulations simulating the Gaussian distribution, calculates the signal ratio histogram, a threshold value Th based on the model It may be calculated and set.

  Next, the temporary phase calculation processing by the temporary phase distribution calculation unit 332 in step S1102 will be described. The temporary phase distribution calculation unit 332 of the present embodiment selects an arbitrary seed point from the seed point candidate pixel group, performs phase unwrapping on the phase value of the ratio image I from the selected seed point by a region enlargement method, The phase value of each pixel is determined, and a temporary phase distribution is calculated. In order to realize this, the provisional phase distribution calculation unit 332 according to the present embodiment performs a phase unwrap process using a region expansion method using a seed point as a reference pixel, as shown in FIG. A phase unwrapping unit 421 for calculation is provided.

  In the phase unwrap process, a phase jump of ± π due to a phase difference between water and fat between adjacent pixels and a phase jump exceeding 2π due to folding are determined. Then, + π or −π is added to the phase of the pixel adjacent to the seed point, and + 2π or −2π is added to make smooth correction. The phase unwrapping unit 421 of the present embodiment calculates the phase difference Δp between the seed point and the adjacent pixel, and calculates + π, −π, + 2π, or −2π of the adjacent pixel according to the value of the phase difference Δp. By adding to the phase, phase unwrap processing is performed.

  In addition, the phase unwrap process of this embodiment is not restricted to the above-mentioned method. The phase may be unwrapped by a known method.

  An outline of the flow of the temporary phase distribution calculation process by the temporary phase distribution calculation unit 332 in step S1102 will be described. FIG. 11 is a flowchart for explaining the flow of the provisional phase distribution calculation process of the present embodiment.

  First, the provisional phase distribution calculation unit 332 sets one arbitrary pixel of the seed point candidate pixel group described above as a seed point in the ratio image I (step S1301). Here, pixels randomly selected from the seed point candidate pixel group are set as seed points. The seed point to be set may be a pixel having the maximum absolute ratio image | I | in the seed point candidate pixel group.

  Next, the phase unwrapping unit 421 performs a phase unwrapping process using the seed point as a reference pixel, and calculates a temporary phase distribution (step S1302). Specifically, the phase difference between the phase of the seed point and the phase of the pixel adjacent to the seed point is calculated. Then, the above-described phase unwrapping process is performed so that this phase difference becomes small. When there are a plurality of adjacent pixels, the phase unwrapping process is performed for all the adjacent pixels.

  Next, the provisional phase distribution calculation unit 332 sets the pixel after the phase unwrapping process as a new seed point (step S1303), expands the processing target area until the processing of all the pixels in the image is completed, The phase unwrapping process is repeated (step S1304) to obtain a temporary phase distribution.

  In this embodiment, the temporary phase distribution is calculated by selecting a seed point from the seed point candidate pixel group and performing phase unwrapping using the region enlargement method, but without using the seed point. The provisional phase distribution may be calculated using a known method for calculating the phase distribution.

  Next, details of the correctness determination processing by the correctness determination unit 333 in step S1103 will be described. The correctness determination unit 333 of this embodiment determines whether the temporary phase distribution calculated by the temporary phase distribution calculation unit 332 is a correct phase distribution or an incorrect phase distribution whose phase value is shifted by π. The determination is performed using the signal of the seed point candidate pixel group specified by the seed point candidate specifying unit 331 of the ratio image I.

仮の位相分布における位置ｒの画素の位相値をθ（ｒ）、φ（ｒ）とθ（ｒ）との位相差をδ（ｒ）、比画像の信号Ｉ（ｒ）の位相からθ（ｒ）を除去した後の比画像の信号（除去後信号値）をＩ’（ｒ）とすると、Ｉ’（ｒ）は、以下の式（１２）で表される。

仮の位相分布θが正しく算出されているとき、すなわち、仮の位相分布の正誤判定が真のときは、δ（ｒ）は０となる。従って、式（１２）は、以下の式（１３）で表される。

一方、仮の位相分布が誤って算出されているとき、すなわち、仮の位相分布の正誤判定が偽のときは、水と脂肪の位相差によってδ（ｒ）はπとなる。従って、式（１２）は、以下の式（１４）で表される。

The position of an arbitrary pixel in the seed point candidate pixel group is r, the water signal intensity at the position r is W (r), the fat signal intensity is F (r), and the time difference dt between the echo times t _in and t _out If the phase change due to the non-uniformity of the static magnetic field generated in (2) is φ (r), the signal I (r) of the ratio image is expressed by the following equation (11).

In the provisional phase distribution, the phase value of the pixel at position r is θ (r), the phase difference between φ (r) and θ (r) is δ (r), and the phase of the signal I (r) of the ratio image is θ ( If the ratio image signal (removed signal value) after removing r) is I ′ (r), I ′ (r) is expressed by the following equation (12).

When the provisional phase distribution θ is correctly calculated, that is, when the correct / incorrect determination of the provisional phase distribution is true, δ (r) is zero. Therefore, Formula (12) is represented by the following Formula (13).

On the other hand, when the temporary phase distribution is erroneously calculated, that is, when the correctness / incorrectness determination of the temporary phase distribution is false, δ (r) becomes π due to the phase difference between water and fat. Therefore, Formula (12) is represented by the following Formula (14).

Therefore, from the equations (13) and (14), when W (r)> F (r), that is, when the main component of the signal value of the pixel at the position r is water, I ′ _true (r)> 0 , I ′ _false (r) <0. When W (r) <F (r), that is, when the main component of the signal value of the pixel at the position r is a fat signal, I ′ _true (r) <0, I ′ _false (r)> 0 Become. In this embodiment, when W (r) = F (r), it is excluded from the calculation target.

  As described above, the number of water pixels included in the seed point candidate pixel group is larger than the number of fat pixels. Therefore, if the provisional phase distribution is correctly calculated, the number of pixels satisfying I ′ (r)> 0 should be larger than the number of pixels satisfying I ′ (r) <0. The correctness determination unit 333 of the present embodiment uses this characteristic to determine the correctness of the temporary phase distribution.

The correctness determination unit 333 calculates the number of pixels N _p satisfying I ′ (r)> 0 and the number of pixels N _n satisfying I ′ (r) <0 in the seed point candidate pixel group of the ratio image I. calculate. When N _p > N _n , the correctness determination result is true, that is, the provisional phase distribution is calculated correctly. On the other hand, when N _p ≦ N _n , it is assumed that the correctness / incorrectness determination result is false, that is, the provisional phase distribution is erroneously output.

In order to perform correct / incorrect determination according to the above procedure, the correctness / incorrectness determination unit 333 of the present embodiment, as shown in FIG. A phase removal unit 431 that removes a phase value calculated as a distribution, and a pixel whose phase-removed signal value (removed signal value) is a positive value for each pixel of the seed point candidate pixel group of the ratio image I the number n _p, compares the values of pixel number finding section 432 that calculates the number of pixels n _n is a negative value, the positive number of pixels is a value n _p and the pixel number n _n is a negative value And a determination unit 433 for determining whether the temporary phase distribution is correct (true or false).

  The true phase distribution determination unit 334 determines the temporary phase distribution as the true phase distribution when the determination result of the correctness / incorrectness determination unit 333 is positive (true). When the correct / incorrect determination result is false (false), a value obtained by adding the phase value π to the temporary phase distribution is determined as the true phase distribution.

  The flow of the process of determining whether the temporary phase distribution is correct and determining the true phase distribution (true phase distribution determining process) will be described with reference to the flowchart of FIG. FIG. 12 is a flowchart of the true phase distribution determination process of the present embodiment.

  First, the phase removal unit 431 removes the phase value of the provisional phase distribution from the signal of the seed point candidate pixel group of the ratio image I (ratio image signal I) (step S1401), and calculates the ratio image signal I ′. To do. Thereby, the phase change due to the influence of the static magnetic field inhomogeneity is removed.

Next, the pixel number calculation unit 432 calculates the number N _{p of} pixels satisfying I ′> 0 and the number N _{n of} pixels satisfying I ′ <0 in the seed point candidate pixel group (step S1402). . Note that pixels satisfying I ′ = 0 are excluded from the calculation targets. Thereafter, the determination unit 433 compares the sizes of the pixel numbers N _p and N _n (step S1403).

When the determination result is N _p > N _n , the true phase distribution determining unit 334 determines the temporary phase distribution as the true phase distribution (step S1405). On the other hand, when the determination result is N _p ≦ N _n , the true phase distribution determination unit 334 adds the phase value π to the temporary phase distribution (step S1404), and sets the phase distribution after the addition as the true phase distribution. Determination is made (step S1405).

  According to the above procedure, the phase distribution resulting from the static magnetic field inhomogeneity can be accurately calculated. That is, even if the seed point is set to a pixel whose main component is fat, and the offset of the phase value π due to the chemical shift of fat is added to the calculated phase distribution, it can be determined. And a phase distribution excluding the offset can be obtained.

Next, the water-fat image calculation method by the water-fat separation unit 340 in step S1004 will be described. The water-fat separation unit 340 uses the two original images I _in and I _out having different echo times obtained in step S1002 and the true phase distribution calculated in step S1003 to obtain a water image and a fat image. calculate. Specifically, the original after dividing each (exp (-iθ)) in the phase value of each pixel of the true phase distribution phase value of each pixel of the image I _out, or, for each pixel of the original image I _in phase After multiplying the value by the phase value of each pixel of the true phase distribution (exp (+ iθ)), a water image and a fat image are calculated by Expression (2-1) and Expression (2-2).

  As described above, the MRI apparatus 100 of the present embodiment includes a magnetostatic generating magnet 102 that generates a static magnetic field in a space where the subject 101 is placed, a transmission unit that transmits a high-frequency magnetic field pulse to the subject 101, and A receiving unit that receives a nuclear magnetic resonance signal generated from the subject by transmitting a high-frequency magnetic field pulse; a gradient magnetic field applying unit that applies a gradient magnetic field for adding position information to the nuclear magnetic resonance signal; and the transmitting unit, The MRI apparatus 100 includes: a computer 109 that controls operations of the gradient magnetic field applying unit and the receiving unit and performs arithmetic processing on the received nuclear magnetic resonance signal. The measurement unit 310 that obtains the nuclear magnetic resonance signals of two different echo times according to the pulse sequence, and the two different magnetic resonance signals from the two different magnetic resonance signals A reconstructing unit 320 for reconstructing an original image with a co-time; a phase distribution calculating unit 330 for calculating a phase distribution that is a phase distribution that changes due to static magnetic field inhomogeneity from the original images with two different echo times; A water fat separation unit 340 that separates a water signal and a fat signal using the original images of two different echo times and the phase distribution, and the phase distribution calculation unit 330 has the two different echo times. Using the original image, a seed point candidate specifying unit 331 that specifies a pixel group affected by noise as a seed point candidate pixel group, the original images of the two different echo times, and information on the seed point candidate pixel group The provisional phase distribution calculation unit 332 that calculates the provisional phase distribution that is the provisional phase distribution, the correctness determination unit 333 that determines whether the temporary phase distribution is correct, and the determination by the correctness determination unit 333. Based on the results provided the true phase distribution determiner 334 to determine the true phase distribution, a.

  The seed point candidate specifying unit 331 includes an absolute value ratio image calculating unit 411 that calculates an absolute value of a ratio image obtained as a ratio of the original images of the two different echo times as an absolute value ratio image, and the seed point A threshold setting unit 412 that sets at least one threshold for specifying a candidate pixel group; and a pixel group specifying unit 413 that specifies the seed point candidate pixel group in the absolute value ratio image using the threshold. May be.

  The threshold setting unit 412 may set the threshold using a distribution of pixel values of the absolute value ratio image. Further, the threshold setting unit 412 may calculate a signal ratio histogram model based on the signal-to-noise ratio of the original images of the two different echo times, and set the threshold based on the model. Further, the threshold value may be set to a value equal to or larger than a value obtained by adding a standard deviation of the pixel value to the mode value of the pixel value of the absolute value ratio image. The threshold value may be set to a value larger than the mode value of the pixel value of the absolute value ratio image. Further, the threshold value may be set to a value of 1.0 or more.

  The temporary phase distribution calculating unit 332 includes a phase unwrapping unit 421 that performs a phase unwrapping process using a region enlargement method from a predetermined reference pixel on the phase value of the ratio image, and calculates the temporary phase distribution. The reference pixel may be selected from the seed point candidate pixel group.

The correctness determination unit 333 removes the phase value obtained as the temporary phase distribution from each signal of the seed point candidate pixel group of the ratio image, and removes the signal value of each pixel of the seed point candidate pixel group. And a pixel number calculation unit 432 for calculating the number of pixels having a positive signal value and the number of pixels having a negative value in each of the seed point candidate pixel groups. And a determination unit 433 that compares the number of pixels having a positive value with the number of pixels having a negative value to determine whether the temporary phase distribution is correct or incorrect. The determination unit 433 may determine that the number of positive pixels is greater than the number of negative pixels.
Further, the true phase distribution determination unit 334 corrects the temporary phase distribution when the determination result of the determination unit 433 is incorrect, determines the corrected phase distribution as the true phase distribution, and determines the determination result. When the nucleation result is positive, the temporary phase distribution may be determined as the true phase distribution.

  In addition, the water fat separation method of the present embodiment uses a water signal from an original image of two different echo times reconstructed from two different echo time nuclear magnetic resonance signals measured according to a predetermined pulse sequence. Water fat separation method for separating a fat signal and a fat signal, using the original images of the two different echo times, specifying a pixel group affected by noise as a seed point candidate pixel group, and the two different echoes Using a temporal original image and information on the seed point candidate pixel group, calculate a temporary phase distribution that is the temporary phase distribution, determine whether the temporary phase distribution is correct, and based on a determination result by the correctness determination unit The true phase distribution is determined, and the water signal and the fat signal are separated using the original images of the two different echo times and the true phase distribution.

  As described above, according to the present embodiment, the phase distribution that is a phase change caused by the static magnetic field inhomogeneity is obtained by using the region enlargement method for the phase value of the ratio image of the original images at two different echo times. Calculation is performed by performing unwrap processing. Prior to the calculation, a seed point candidate pixel group is specified so that the probability that the seed point, which is the starting point of the phase unwrapping process, is a pixel whose water signal is the main pixel is increased. Therefore, the possibility that the offset of the phase value π is added to the phase distribution calculated by the phase unwrapping process is reduced. That is, the possibility that an accurate phase distribution is calculated increases.

  Further, the correctness / incorrectness of the calculated phase distribution is determined. At this time, determination is performed based on the number of pixels using a seed point candidate pixel group having a high probability that the water signal is the main pixel. Therefore, highly accurate determination can be easily performed. Since the phase distribution is finally determined based on the highly accurate determination result, the possibility that the accurate phase distribution is calculated further increases. Then, according to the present embodiment, the water signal and the fat signal are separated by using the phase distribution that is highly likely to be an accurate phase distribution, and a water image and a fat image are obtained.

  Therefore, according to the present embodiment, the possibility that the obtained water image and fat image are correct is extremely high. That is, according to the present embodiment, the water image and the fat image are improved by increasing the accuracy of correctly calculating the phase image in which the static magnetic field inhomogeneity is reflected in the separation photography of water and fat using the 2-Point Dixon method. Can be separated and water and fat can be separated with high accuracy without burdening the operator.

In the above embodiment, is used as the two echo times different for acquiring echo signals, the echo time t _in the water and the fat the same phase, and the echo time t _out of the water and the fat phase opposition Although the case has been described as an example, the acquisition time of the echo signal is not limited to this. For example, the two echo times may be TE ₁ and TE ₂ that satisfy the following equation (15).
cos (2πf _FW TE ₁ ) ≠ cos (2πf _FW TE ₂ ) (15)
f _FW is the frequency difference between water and fat.

In this case, δ _W and δ _F are used instead of the ratio image when calculating the temporary phase distribution in S1102. Note that δ _W is a phase distribution calculated when all the tissue signals are assumed to be water, and δ _F is a phase distribution calculated when all the tissue signals are assumed to be fat. Then, when performing phase unwrapping by the enlarged region method, the phase of the pixel adjacent to the seed point is selected from the δ _W phase and δ _F phase of the pixel that is closer to the phase of the seed point pixel. To do.

  DESCRIPTION OF SYMBOLS 100: MRI apparatus, 101: Subject, 102: Static magnetic field coil, 103: Gradient magnetic field coil, 104: Shim coil, 105: Transmission coil, 106: Reception coil, 107: Transmitter, 108: Receiver, 109: Calculator, 110: Display, 111: External storage device, 112: Power supply unit for gradient magnetic field, 113: Power supply unit for shim, 114: Sequence control device, 115: Input device, 120: MRI device, 130: MRI device, 310: Measurement unit 320: Reconstruction unit 330: Phase distribution calculation unit 331: Seed point candidate identification unit 332: Temporary phase distribution calculation unit 333: Correct / incorrect determination unit 334: True phase distribution determination unit 340: Water-fat separation Unit, 411: absolute value ratio image calculation unit, 412: threshold value setting unit, 413: pixel group specifying unit, 421: phase unwrapping unit, 431: phase removing unit, 32: Pixel number calculation unit, 433: determination unit, 510: pulse sequence, 511: RF pulse, 512: slice selection gradient magnetic field, 513: phase encoding gradient magnetic field, 514: readout gradient magnetic field, 515: first echo, 516: Read gradient magnetic field, 517: second echo, 520: pulse sequence, 521: RF pulse, 522: slice selective gradient magnetic field, 523: phase encode gradient magnetic field, 524: read gradient magnetic field, 525: first echo, 526: read gradient Magnetic field, 527: second echo, 528: RF pulse, 529: slice selective gradient magnetic field, 601: water signal ratio distribution, 602: fat signal ratio distribution, 603: threshold Th

Claims (12)

A magnetostatic generating magnet that generates a static magnetic field in a space in which the subject is placed, a transmission unit that transmits a high-frequency magnetic field pulse to the subject, and a nuclear magnetic resonance signal generated from the subject by transmission of the high-frequency magnetic field pulse The receiving unit, the gradient magnetic field applying unit for applying a gradient magnetic field for adding position information to the nuclear magnetic resonance signal, the operation of the transmitting unit, the gradient magnetic field applying unit, and the receiving unit are controlled and the reception is performed. A magnetic resonance imaging apparatus comprising: a computer that performs arithmetic processing on a nuclear magnetic resonance signal;
The calculator is
A measurement unit for obtaining the nuclear magnetic resonance signals of two different echo times according to a predetermined pulse sequence;
A reconstruction unit for reconstructing the original images of the two different echo times from the two nuclear magnetic resonance signals;
A phase distribution calculating unit that calculates a phase distribution that is a phase distribution that changes due to non-uniform static magnetic field from the original images of two different echo times;
A water fat separation unit that separates a water signal and a fat signal using the original images of the two different echo times and the phase distribution;
The phase distribution calculator is
Using the original images of the two different echo times, a seed point candidate specifying unit for specifying a pixel group affected by noise as a seed point candidate pixel group;
A temporary phase distribution calculation unit that calculates a temporary phase distribution that is the temporary phase distribution using the original images of the two different echo times and the information of the seed point candidate pixel group;
A correct / incorrect determination unit for determining correctness of the provisional phase distribution;
E Bei and a true phase distribution determination section for determining the true phase distribution based on the determination result by the accuracy determination unit,
The seed point candidate specifying unit
An absolute value ratio image calculation unit that calculates an absolute value of a ratio image obtained as a ratio of original images of two different echo times as an absolute value ratio image;
A threshold setting unit that sets at least one threshold for specifying the seed point candidate pixel group;
And a pixel group specifying unit that specifies the seed point candidate pixel group in the absolute value ratio image using the threshold value . The magnetic resonance imaging apparatus according to claim 1 ,
The magnetic resonance imaging apparatus, wherein the threshold setting unit sets the threshold using a distribution of pixel values of the absolute value ratio image. The magnetic resonance imaging apparatus according to claim 1 ,
The magnetic resonance imaging apparatus, wherein the threshold setting unit calculates a signal ratio histogram based on a signal-to-noise ratio of the original images of the two different echo times and sets the threshold based on the model. The magnetic resonance imaging apparatus according to claim 2 ,
The magnetic resonance imaging apparatus, wherein the threshold value is set to a value equal to or greater than a value obtained by adding a standard deviation of the pixel value to the mode value of the pixel value of the absolute value ratio image. The magnetic resonance imaging apparatus according to claim 2 ,
The magnetic resonance imaging apparatus, wherein the threshold value is set to a value larger than a mode value of pixel values of the absolute value ratio image. The magnetic resonance imaging apparatus according to claim 1 ,
The threshold value is set to a value of 1.0 or more. The magnetic resonance imaging apparatus according to claim 1 ,
The provisional phase distribution calculator is
A phase unwrapping unit that performs a phase unwrapping process using a region enlargement method from a predetermined reference pixel on the phase value of the ratio image, and includes a phase unwrapping unit that calculates the temporary phase distribution,
The magnetic resonance imaging apparatus, wherein the reference pixel is selected from the seed point candidate pixel group. The magnetic resonance imaging apparatus according to claim 1 ,
The correctness determination unit
A phase removal unit that removes each phase value obtained as the temporary phase distribution from each signal of the seed point candidate pixel group of the ratio image and obtains a signal value after removal of each pixel of the seed point candidate pixel group;
In each seed point candidate pixel group, a pixel number calculation unit for calculating the number of pixels where the signal value after removal is a positive value and the number of pixels where the signal value is a negative value,
A magnetic resonance imaging apparatus comprising: a determination unit that compares the number of pixels having a positive value with the number of pixels having a negative value to determine whether the temporary phase distribution is correct or incorrect. The magnetic resonance imaging apparatus according to claim 8 ,
The determination unit determines to be positive when the number of pixels having a positive value is greater than the number of pixels having a negative value. The magnetic resonance imaging apparatus according to claim 8 ,
The true phase distribution determining unit
When the determination result of the determination unit is incorrect, the temporary phase distribution is corrected, the corrected phase distribution is determined as the true phase distribution, and when the determination result nucleation result is positive, the temporary phase distribution Is determined to be the true phase distribution. A water / fat separation method for separating a water signal and a fat signal from the original images of two different echo times reconstructed from nuclear magnetic resonance signals of two different echo times measured according to a predetermined pulse sequence, respectively. There,
An absolute value of a ratio image obtained as a ratio of the original images of two different echo times is calculated as an absolute value ratio image, and at least a threshold value for specifying a pixel group affected by noise as a seed point candidate pixel group One is set and the seed point candidate pixel group in the absolute value ratio image is specified using the threshold value ,
Using the original images of the two different echo times and the information of the seed point candidate pixel group, calculate a temporary phase distribution that is the temporary phase distribution,
Determine whether the temporary phase distribution is correct or incorrect,
Determine the true phase distribution based on the determination result by the correctness determination unit,
A water fat separation method characterized by separating a water signal and a fat signal using the original images of the two different echo times and the true phase distribution.   A magnetostatic generating magnet that generates a static magnetic field in a space in which the subject is placed, a transmission unit that transmits a high-frequency magnetic field pulse to the subject, and a nuclear magnetic resonance signal generated from the subject by transmission of the high-frequency magnetic field pulse The receiving unit, the gradient magnetic field applying unit for applying a gradient magnetic field for adding position information to the nuclear magnetic resonance signal, the operation of the transmitting unit, the gradient magnetic field applying unit, and the receiving unit are controlled and the reception is performed. A magnetic resonance imaging apparatus comprising: a computer that performs arithmetic processing on a nuclear magnetic resonance signal;
  The calculator is
  A measurement unit for obtaining the nuclear magnetic resonance signals of two different echo times according to a predetermined pulse sequence;
  A reconstruction unit for reconstructing the original images of the two different echo times from the two nuclear magnetic resonance signals;
  A phase distribution calculating unit that calculates a phase distribution that is a phase distribution that changes due to non-uniform static magnetic field from the original images of two different echo times;
  A water fat separation unit that separates a water signal and a fat signal using the original images of the two different echo times and the phase distribution;
  The phase distribution calculator is
  Using the original images of the two different echo times, a seed point candidate specifying unit for specifying a pixel group affected by noise as a seed point candidate pixel group;
  Using the original images of the two different echo times, a temporary phase distribution calculating unit that calculates a temporary phase distribution that is the temporary phase distribution;
  Using the information of the seed point candidate pixel group, a correct / incorrect determination unit that determines whether the temporary phase distribution is correct,
  A true phase distribution determination unit that determines a true phase distribution based on a determination result by the correctness determination unit,
  The seed point candidate specifying unit
  An absolute value ratio image calculation unit that calculates an absolute value of a ratio image obtained as a ratio of original images of two different echo times as an absolute value ratio image;
  A threshold setting unit that sets at least one threshold for specifying the seed point candidate pixel group;
  A pixel group specifying unit that specifies the seed point candidate pixel group in the absolute value ratio image using the threshold value.
  A magnetic resonance imaging apparatus.

Images