Magnetic Resonance Imaging Apparatus, Image Processing Apparatus, and Phase Correction Method

Abstract

The present invention relates to a magnetic resonance imaging apparatus, an image processing apparatus, and a phase correction method. The following technique is provided: in imaging using EPI, generation of artifacts during phase correction for each channel is avoided, and phase correction is performed with good accuracy. Pre-scan data for each channel is used to calculate a common phase correction value applied to data in all channels. The common phase correction value is obtained by synthesizing differential phases obtained for each channel. The differential phase is obtained by complex accumulation while maintaining the absolute value of each channel unchanged. The synthesis is performed by complex averaging, and averaging processing corresponding to the weight of the absolute value is performed. By using the common phase correction value, generation of artifacts can be prevented. In addition, by including the weight of the absolute value, robust phase correction can be performed.

Description

Technical Field

[0001] The present invention relates to a magnetic resonance imaging apparatus (hereinafter abbreviated as an MRI apparatus), and particularly to a phase correction technique for imaging based on the EPI method using a plurality of receiver coils. Background Art

[0002] As one of the high-speed imaging methods using an MRI apparatus, there is EPI (Echo Planar Imaging), which is used in various imaging such as imaging for obtaining a diffusion-weighted image. Since EPI can acquire data in the k-space (Fourier space) where nuclear magnetic resonance signals (NMR signals) are arranged with a single excitation, it has the advantage of being able to acquire signals at high speed. However, there are still problems such as: when acquiring an NMR signal as an echo signal while reversing the readout gradient magnetic field, N / 2 artifacts are generated due to the application error of the gradient magnetic field, distortion is likely to occur due to inhomogeneity of the static magnetic field, and position deviation in the phase encoding direction.

[0003] Regarding these problems, various N / 2 artifact correction methods and phase correction techniques have been proposed. For example, there are the following methods (Patent Document 1): correcting the phase error in the x-ky space based on the data obtained by pre-scanning to correct the position deviation in the kx-y direction. According to this method, reduction of N / 2 artifacts and position deviation in the y direction can be achieved.

[0004] On the other hand, as one of the high-speed imaging methods in MRI, there is the parallel imaging (PI) method. In this parallel imaging method, a plurality of receiver coils are used, signals are acquired by omitting data in the phase encoding direction of the k-space, and using the sensitivity distribution of these receiver coils, unmeasured data is estimated by calculation to obtain an image without aliasing. In this PI method, EPI can be used as a signal acquisition method, thereby reducing distortion and blurring. In this case, it is also necessary to correct the phase error caused by EPI. However, when calculating the phase correction coefficient (correction amount) for each receiver coil (also called a channel) to perform phase correction and synthesizing the phase-corrected data, an error occurs in the phase correction amount due to differences in SN, etc. for each channel, and aliasing artifacts may occur in the synthesized image. In the PI method, since the sensitivity distribution of the receiver coils is used, a deviation between the sensitivity distribution and the phase-corrected data may also cause aliasing artifacts.

[0005] The following techniques are disclosed in Patent Documents 2 and 3: combining the PI method and EPI, and performing phase correction for eliminating position deviation and the like caused by EPI. In the technique disclosed in Patent Document 2, when obtaining phase correction coefficients based on pre-scan data of multiple channels of a receiving coil, in order to solve the problem that sufficient SNR cannot be obtained for each channel, NMR signals received by a body coil separately installed from a multi-channel surface coil are used to calculate the phase correction coefficients of the surface coil, thereby correcting the NMR signals received by the surface coil. In addition, in the technique disclosed in Patent Document 3, first k-space data and second k-space data with different rows in k-space are used to generate intermediate images (first aliased image and second aliased image) of a multi-channel quantity, and based on this, a phase difference map is created to perform phase correction on the image.

[0006] Prior Art Documents

[0007] Patent Documents

[0008] Patent Document 1: Japanese Unexamined Patent Application Publication No. 2019-162370

[0009] Patent Document 2: Japanese Patent No. 6348355 Specification

[0010] Patent Document 3: Japanese Patent No. 6283134 Specification

[0011] The technique described in Patent Document 2 requires obtaining additional signals through a body coil, and when sufficient signals cannot be obtained with the body coil, the accuracy of phase correction may be reduced. The technique described in Patent Document 3 requires creating intermediate images, resulting in a large computational burden. In addition, since a phase difference map is obtained for each channel, when performing phase correction on each data of the channel using the phase difference map of each channel, the problem of aliasing artifacts cannot be solved. Summary of the Invention

[0012] The subject of the present invention is to provide the following technique: in imaging that combines the PI method and EPI, phase correction can be performed with good data accuracy for each channel without the need for phase correction for each channel.

[0013] To solve the above subject, the present invention uses pre-scan data of each channel to calculate a common phase correction value applied to data of all channels. The common phase correction amount is obtained by synthesizing the differential phases obtained for each channel. At this time, the differential phase is performed while maintaining the signal values (absolute values) of each channel unchanged, and averaging processing corresponding to the weights of the absolute values is performed.

[0014] That is, the MRI apparatus of the present invention includes: an imaging unit that applies a high-frequency magnetic field pulse and an inclined magnetic field pulse according to a given pulse sequence, and collects nuclear magnetic resonance signals generated from an object to be examined; and an arithmetic unit that uses the nuclear magnetic resonance signals to create an image of the object to be examined. The imaging unit includes a multi-channel receiving coil, and as the pulse sequence, an EPI method pulse sequence that acquires a plurality of nuclear magnetic resonance signals after applying an excitation high-frequency magnetic field pulse once is executed. The arithmetic unit includes a preprocessing unit that corrects the phase of data composed of nuclear magnetic resonance signals respectively obtained in a plurality of channels of the receiving coil by executing the pulse sequence of the EPI method. The preprocessing unit includes: a correction value calculation unit that calculates a phase correction value using correction data respectively received in a plurality of channels, and synthesizes the phase correction values of each channel to calculate a common phase correction value; and a phase correction unit that corrects the phase of image formation data received in each channel using the common phase correction value.

[0015] In addition, in the phase correction method of the present invention, in a magnetic resonance imaging apparatus having a multi-channel receiving coil, the phase of image formation data respectively acquired in a plurality of channels by an EPI sequence is corrected. The phase correction method includes: a step of calculating a phase correction value using correction data respectively received in a plurality of channels; a step of synthesizing the phase correction values of each channel to calculate a common phase correction value; and a step of correcting the phase of image formation data received in each channel using the common phase correction value. The step of calculating the phase correction value includes a step of calculating the differential phase between rows in the x-ky space data of the correction data by complex accumulation, and calculating the phase correction value based on the differential phase. In addition, the step of calculating the common phase correction value includes a step of performing cumulative averaging on the phase correction values of each channel.

[0016] Advantages of the Invention

[0017] According to the present invention, by correcting the phase of the data of each receiving coil using a common phase correction value, it is possible to avoid wrap-around artifacts caused by different phase correction values for each channel, and to provide an MR image without position deviation and artifacts. In addition, by calculating the common phase correction value by weighting with the signal values (absolute values) of the data of each receiving coil, robust correction can be performed with good accuracy. Furthermore, it is possible to prevent division by zero caused by division operations during differential calculation, and to simply and stably achieve channel synthesis. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a diagram showing the overall structure of the MRI apparatus.

[0019] Figure 2 It is a diagram showing an example of the EPI sequence.

[0020] Figure 3 It is a diagram showing a method for collecting k-space data based on EPI. (A) is an example of single-shot EPI, (B) is an example of multi-shot EPI, and (C) is a case of performing single-shot EPI twice.

[0021] Figure 4 It is a functional block diagram showing the functions of a computer.

[0022] Figure 5 It is a diagram showing the flow of the phase correction process in the arithmetic unit.

[0023] Figure 6 It is a functional block diagram showing the functions of the preprocessing unit in Embodiment 1.

[0024] Figure 7 (A) to (C) of Figure 6 are diagrams showing the details of the x-direction inclination calculation unit, the x-direction intercept calculation unit, and the ky-direction inclination calculation unit of

[0025] Figure 8 It is a diagram showing the processing flow of the x-direction correction in Embodiment 1.

[0026] Figure 9 It is a diagram for explaining the calculation of the differential phase.

[0027] Figure 10 It is a diagram showing an example of the kernel function used in the slice synthesis of the phase correction value.

[0028] Figure 11 It is a diagram for explaining the calculation of the inclination and intercept in the x direction.

[0029] Figure 12 It is a diagram showing the processing flow of the ky-direction correction in Embodiment 1.

[0030] Figure 13 (A) to (C) of

[0031] Figure 14 are diagrams showing the phase diagrams of the x-ky space data. (A) is the data of each excitation, (B) is the data after being segmented into patterns, and (C) represents the data after the phase differentiation process.

[0032] Figure 15 It is a diagram for explaining the difference in effects between (A) of the present embodiment and (B) of the comparative example.

[0033] Figure 16 It is a diagram showing the overall structure of the image processing apparatus.

[0034] Description of reference numerals

[0035] 1: MRI device, 2: Image processing device, 10: Imaging unit, 20: Computer, 30: UI unit, 31: Display device, 32: Input device, 40: Storage device, 21: Measurement control unit, 22: Display control unit, 23: Preprocessing unit, 24: Image reconstruction unit, 231: Correction data creation unit, 233: Phase correction unit, 51: X-direction correction unit, 52: Ky-direction correction unit, 511: Differential phase calculation unit, 512: Inter-channel averaging unit, 513: X-direction inclination calculation unit, 514: X-direction intercept calculation unit, 515: Phase correction unit, 521: Data segmentation unit, 522: Ky-direction inclination calculation unit, 523: Data unification unit, 524: Phase correction unit Detailed implementation mode

[0036] The following describes the implementation modes of the MRI device and the image processing method of the present invention.

[0037] First, refer to Figure 1 to describe the overall structure of the MRI device 1 of the present invention. As Figure 1 shown, the MRI device includes an imaging unit 10, a computer 20, a UI (user interface) unit 30, and a storage device 40. The UI unit 30 is used for the user to input instructions required for the operation of the MRI device, or to display images captured by the MRI device, required UIs for input, etc., and includes a display device 31, an input device 32, etc. The storage device 40 can be a removable medium directly connected to the MRI device, an HDD, etc., or a server connected through a network, etc., or a storage medium in the cloud.

[0038] The imaging unit 10 includes: a static magnetic field magnet (magnet) 11 that generates a static magnetic field in the space (imaging space) where the subject is placed; a gradient magnetic field coil 12 that applies a magnetic field gradient to the static magnetic field; and a probe 13 that applies a high-frequency magnetic field to the subject and detects the NMR signal generated from the subject.

[0039] The static magnetic field magnet 11 has a permanent magnet method, an electromagnet method, a superconducting method, etc. according to the method of generating the static magnetic field. In addition, according to the orientation of the static magnetic field in the imaging space, there are a vertical magnetic field method and a horizontal magnetic field method. The present invention can correspond to any of these methods.

[0040] The gradient magnetic field coil 12 includes three sets of coils that generate gradient magnetic fields in three mutually orthogonal directions of X, Y, and Z respectively, and are respectively connected to the gradient magnetic field power supply 14. A gradient magnetic field pulse of a desired magnitude is generated by the current supplied from the gradient magnetic field power supply 14. By combining the currents flowing through the coils of the three axes, a gradient magnetic field in an arbitrary direction can be generated in the imaging space. Usually, gradient magnetic field pulses in the three-axis directions of the slice direction, the phase encoding direction, and the readout direction (also referred to as the frequency encoding direction) are generated, thereby adding position information to the NMR signal. In addition, a shim coil for correcting the inhomogeneity of the static magnetic field is sometimes provided. Furthermore, the gradient magnetic field coil 12 sometimes also serves as a shim coil.

[0041] Although not shown in the figure, the probe 13 includes a transmit RF coil that generates a high-frequency magnetic field pulse (referred to as an RF pulse) and a receive RF coil that receives the NMR signal generated from the subject. The transmit RF coil is connected to the high-frequency magnetic field generator 15, and the receive RF coil is connected to the receiver 16. They can be served by one coil for both transmission and reception, or sometimes separate coils are prepared. In the case where one coil serves for both transmission and reception, a switch (not shown) is inserted between the high-frequency magnetic field generator 15 and the receiver 16. In the present embodiment, at least the coil that functions as a receiver is a multi-channel coil formed by arranging a plurality of coils, and is respectively connected to the receiver 16. The receiver 16 includes a receive amplifier, an orthogonal detection circuit, an A / D converter, etc., and sends the digitized signal to the computer 20.

[0042] The imaging unit 10 further includes a gradient magnetic field power supply 14, a high-frequency magnetic field generator 15, and a sequence generator 17 that causes the receiver 16 to operate according to a given pulse sequence. During imaging, a pulse sequence determined according to the imaging method is set in the sequence generator 17, and the sequence generator 17 controls the magnitude, application timing of the RF pulse and the gradient magnetic field pulse generated by the transmit RF coil and the gradient magnetic field coil 12, the sampling time of the NMR signal of the receiver 16, etc. according to the set pulse sequence.

[0043] Various pulse sequences different according to the imaging method are previously stored as programs in the memory or storage device 40 in the computer 20. The computer 20 calculates the pulse sequence (imaging sequence) actually used in imaging using the pulse sequence selected corresponding to the imaging method and the imaging parameters set via the UI unit 30, and sets it in the sequence generator 17. In the present embodiment, as a basic pulse sequence, the EPI sequence is executed.

[0044] The EPI sequence is a well-known sequence, and basically as Figure 2As shown, after applying the excitation RF pulse 201 together with the slice selection gradient magnetic field 202, dephasing gradient magnetic fields 203 and 205 are applied in the phase encoding direction and the readout direction, and while reversing the gradient magnetic fields 206 and 207 in the readout direction, multiple echoes 208 are generated. At this time, a blip-shaped phase encoding gradient magnetic field pulse 204 is applied. Thus, as Figure 3 shown in (A), multiple lines of data in the k-space are collected. Figure 2 This is a 2D sequence, but by using phase encoding in two directions, 3D k-space data can also be obtained. In addition, in the EPI sequence, in addition to the single-shot EPI that collects k-space data by one excitation (1 excitation), there is also a multi-shot EPI that collects k-space data by multiple excitations as Figure 3 shown in (B). In addition, although not shown, there is also a multi-shot that divides the k-space into multiple regions for collection. In addition, as Figure 3 shown in (C), there are also cases where single-shot EPI is imaged twice and they are added together. In this case, as Figure 3 shown in (C), generally, the positive and negative polarities (the scanning direction of the k-space) are reversed for imaging. The present invention can be applied to any type of EPI. In addition, the collected k-space data may sometimes be all the data occupying the k-space, or may sometimes be undersampled data. The latter is reconstructed into an image by the operation of the PI method.

[0045] The computer 20 can be composed of a general-purpose computer or workstation equipped with a memory, a CPU, and a GPU, and functions as a control unit for controlling the operations of each part of the imaging unit 10, and also functions as an arithmetic unit for performing various operations for image reconstruction using the NMR signals (k-space data) collected by the imaging unit 10. Among them, a part of the functions of the arithmetic unit can also be implemented by hardware such as a computer, an ASIC, or an FPGA different from the computer 20.

[0046] Refer to Figure 4 the functional block diagram to illustrate the specific functions of the computer 20.

[0047] As shown in the figure, the computer 20 includes: a measurement control unit 21 for controlling the operations of the imaging unit 10; a display control unit 22 for controlling the UI unit 30 (display device 31); a preprocessing unit 23 for preprocessing the measurement data collected by the imaging unit 10; and an image reconstruction unit 24 for reconstructing an image using the preprocessed measurement data. The preprocessing unit 23 and the image reconstruction unit 24 are mainly included in the functions of the arithmetic unit.

[0048] The measurement control unit 21 calculates a pulse sequence used in imaging using imaging parameters such as the doubling rate R (a value that determines the inter-division rate of the PI method), the excitation number, the echo time: TE, and the repetition time: TR set via the UI unit 30, and controls the operation of the imaging unit 10 via the sequence generator 17. In the present embodiment, control is performed to execute measurement (formal scan) for acquiring an image of the subject to be examined and preliminary measurement (pre-scan) for collecting data required for correcting the measurement data obtained in the formal scan. The pulse sequence of the pre-scan is basically the same as that of the formal scan, and is obtained without applying a gradient magnetic field in the phase encoding direction during signal reception.

[0049] The display control unit 22 controls the display mode of the MR image displayed on the display device 31, the display of the GUI for exchanging with the user, and the like.

[0050] The preprocessing unit 23 corrects the measurement data when the imaging unit 10 executes the EPI sequence. In the embodiment of the present invention, it has: a correction data creation unit 231 that prepares data (correction data) for calculating a phase correction value based on the pre-scan data obtained for each channel; and a phase correction unit 233 that calculates a differential phase based on the correction data for each channel, synthesizes them to calculate a common phase correction value for the channels, and uses this phase correction value to perform phase correction on the formal scan data.

[0051] The preprocessing unit 23 performs processing in the x-ky space. Specifically, as Figure 5 shown, the correction data creation unit 231 performs Fourier transform on the pre-scan data (k-space data) in the x direction to obtain x-ky space data, and prepares data for calculating a correction value based on the x-ky space data (S501).

[0052] Next, the phase correction unit 233 calculates correction values for the x direction and the ky direction respectively, and performs phase correction (S502, S503).

[0053] The specific calculation method in the phase correction unit 233 will be described later. In the MRI apparatus of the present embodiment, it is characterized in that, in each of the x-direction phase correction and the ky-direction phase correction, instead of using the phase correction values calculated for each channel of each channel in the phase correction, a common phase correction value is calculated for each channel and used in the phase correction. Thereby, artifacts appearing in the synthesized image due to deviations between channels can be suppressed.

[0054] Furthermore, when calculating the common phase correction value for each channel, the phase correction unit 233 of the present embodiment obtains the differential phase for each channel through complex accumulation, and performs complex averaging on the differential phase of each channel to calculate the phase correction value. In this way, by using complex accumulation as the method for calculating the differential phase, the signal value (absolute value) of each channel can be reflected as a weight in the common phase correction value. As a result, for example, for the correction data with a low signal value where it is difficult to accurately obtain the phase correction value, since the weight in the common phase correction value becomes lower, the accuracy of the common phase correction value is improved. In addition, by using complex accumulation, when obtaining the differential phase, it is possible to prevent division by zero caused by division operations, and stable inter-channel synthesis can be achieved.

[0055] The phase correction unit 233 performs phase correction on the data collected during the formal scan using the phase correction value. At this time, the phase correction in the x direction and the phase correction in the ky direction can be performed independently.

[0056] The image reconstruction unit 24 performs operations such as Fourier transform on the formally scanned data after phase correction to reconstruct the image of the subject (S504). When the formal scan ratio rate is greater than 1, that is, in the case of PI method imaging, the operations of the PI method are used, and undersampling is performed to generate an image that unfolds the aliasing that will occur in the image. Known operations of the PI method include a method of removing aliasing in real space (SENSE), a method of estimating unmeasured data in the measurement space (GRAPPA), etc., and any of them can be adopted. In the present embodiment, the SENSE method is used for illustration.

[0057] In the operations of the SENSE method, the sensitivity distribution of the receive coil is used to remove the aliasing of the image. By performing phase correction before the operations of the SENSE method, the phase deviation between the sensitivity distribution of the receive coil and the phase of the measurement data can be eliminated, and the deterioration of the image quality caused by the phase deviation can be prevented.

[0058] The following describes a specific embodiment of the processing in the preprocessing unit 23.

[0059] <Embodiment 1>

[0060] This embodiment uses the pre-scan data obtained for each channel by single-shot 2D-EPI with 2 accumulations to calculate the common phase correction value. In addition, the phase difference data of adjacent slices is synthesized and used to calculate the common phase correction value, but this is not necessary. In the k-space scan of the single-shot EPI with 2 accumulations in this embodiment, as shown in (C) of Figure 3 , in the first and second times, the application direction of the readout gradient magnetic field is made opposite, that is, the scanning directions in the k-space are made opposite. Therefore, the obtained correction data becomes a set of data with opposite scanning directions.

[0061] The following is for reference Figures 6 - 8 to describe in detail the processing of this embodiment. Figure 6 FIG. is a functional block diagram showing the functions of the preprocessing unit 23 of this embodiment.

[0062] The preprocessing unit 23 of this embodiment is provided with, as Figure 6 shown: an x-direction correction unit 51 that performs phase correction in the readout direction (x-direction); and a ky-direction correction unit 52 that performs phase correction in the phase encoding direction. The x-direction correction mainly corrects the phase difference generated between rows in k-space, which is specific to EPI. The ky-direction correction corrects the phase error generated in the phase encoding direction due to inhomogeneity of the static magnetic field or the like.

[0063] The x-direction correction unit 51 further includes a differential phase calculation unit 511, an inter-channel averaging unit 512, an x-direction slope calculation unit 513, an x-direction intercept calculation unit 514, and a phase correction unit 515. The x-direction slope calculation unit 513 further includes, as Figure 7 shown in (A) of FIG., an x-direction phase differentiation calculation unit and an x-direction averaging calculation unit. The x-direction intercept calculation unit 514 includes, as Figure 7 shown in (B) of FIG., an offset calculation unit and an x-direction averaging calculation unit.

[0064] The ky-direction correction unit 52 is provided with: a data segmentation unit 521 that segments the prescan data used for correction; a ky-direction slope calculation unit 522 that calculates the ky-direction slope for each of the segmented data; a data unification unit 523 that unifies the ky-direction slopes of the segmented data; and a phase correction unit 524. The ky-direction slope calculation unit 522 further includes, as Figure 7 shown in (C) of FIG., a ky-direction phase differentiation calculation unit; and an inter-channel averaging calculation unit that calculates the inter-channel average for the x-direction and the ky-direction.

[0065] Next, the preprocessing flow of this embodiment with the above structure will be described. First, the x-direction correction flow will be described with reference to Figure 8 FIG.. The x-direction correction is as Figure 8include the following as shown: Step S801, perform Fourier transform on the pre-scanned data (k-space data) obtained by the imaging unit 10 in the x direction, and obtain the data for calculating the correction value (x-ky space data) for each channel; Step S802, extract the rows used for correction in the x-ky space data; Step S803, calculate the differential phase through complex accumulation using the data of the extracted rows; Step S804, obtain the inter-channel average of the differential phase calculated in Step S803 through complex averaging; Step S805, synthesize the differentially averaged differential phase between adjacent slices; Step S806, calculate the inclination in the x direction using the synthesized differential phase; Step S807, calculate the intercept in the x direction; and Step S808, perform phase correction on the data obtained for each channel in the formal scan using the inclination and intercept in the x direction obtained in Step S806 and Step S807 respectively (collectively referred to as the phase correction coefficient). The details of the processing of each step are described below.

[0066] [Step S801]

[0067] In this step, the pre-processing unit 23 (correction data creation unit 231) performs Fourier transform on the k-space data obtained by single excitation with accumulation twice in the x direction to obtain the x-ky space data. Since the data is collected in such a way that the scanning directions of the first scan and the second scan are opposite, in the x-ky space data, a phase deviation (phase deviation in the x direction) caused by the deviation of the symmetry of the inclined magnetic field shape occurs between adjacent rows.

[0068] [Step S802]

[0069] The x-direction correction unit 51 corrects the above-mentioned phase deviation in the x direction. Therefore, the rows (rows in the ky direction) used for phase correction in the x direction are extracted from the x-ky space data. The extracted rows are, for example, the rows corresponding to ky = 0 where the signal value becomes the maximum.

[0070] [Step S803]

[0071] The differential phase calculation unit 511 calculates the differential phase through complex accumulation of the first row data and the second row data. If the first row data is set as l1 and the second row data is set as l2, the complex accumulation d characterized by Equation (1) can be used to s calculate the differential phase d through Equation (2).

[0072]

Mathematical formula 1

[0073]

[0074]

Mathematical formula 2

[0075]

[0076] In Equation (1), "l * 2" represents the complex conjugate of "l2". Further, in Equation (1) and (2), x represents the position in the x direction, z represents the position in the slice direction, and ch represents the channel number (the same applies hereinafter). By performing such complex accumulation, the difference is calculated for the phase, and the signal values (absolute values) are accumulated. The signal values that are accumulated and squared take the square root in Equation (2) and return to the first power.

[0077] As Figure 9 shown, by performing complex accumulation on the phase of the first data and the phase of the second data, a first-order component corresponding to the deviation of the positions of the k-space peaks in the positive and negative directions can be obtained. Further, in Equation (1), by performing complex accumulation instead of complex division operation, the differential phase can be calculated while maintaining the signal amount unchanged. Since the absolute value of d corresponds to the signal amount, it becomes a form in which the signal amount is reflected in the reliability of the differential phase.

[0078] [Step S804]

[0079] The inter-channel averaging unit 512 calculates the inter-channel average c of the differential phase calculated in Equation (2). Using Equation (3), the inter-channel average c is calculated by complex averaging. By calculating using complex averaging, the absolute value can be maintained, and weighting based on the absolute value for each channel is also performed here.

[0080] [Mathematical Formula 3]

[0081]

[0082] In Equation (3), "n ch " represents the number of channels.

[0083] [Step S805]

[0084] In 2D imaging, since data is acquired for a given region over multiple slices (z > 1 in Equation (1)), in this case, the differential phases of multiple slices adjacent to each other are synthesized by taking the complex average. The number L of slices to be synthesized is not particularly limited, and is set to several to about 10, for example, determined by Equation (4). In the equation, Z is the thickness of the synthesis target region that can be arbitrarily specified (for example, set to 20 mm), and SliceGap is the distance between adjacent slices.

[0085] [Mathematical Formula 4]

[0086]

[0087] Complex average c mIt is calculated by performing a convolution operation using the kernel function g(k) as represented by the following equation (5). Additionally, the kernel function g(k) is, for example, Figure 10 a Gaussian function (solid line) as represented by, and can be defined by Equation (6).

[0088]

Mathematical formula 5

[0089]

[0090]

Mathematical formula 6

[0091] g(k) = exp(-k 2 / 2σ 2 )(6)

[0092] σ = L / 5

[0093] By using such a Gaussian kernel, the contribution is greater the closer to the vicinity, and the phase correction value can be calculated stably and with high precision. However, the kernel function is not limited to this, and any function can be used corresponding to the number of slices, the interval between slices, etc.

[0094] This process is performed for each slice. As a result, the complex average cm can be obtained for each slice. Additionally, this step is not essential in the calculation of the phase correction value, but by performing slice synthesis, even when there is a reason that causes a slight reduction in precision in one slice, its influence can also be reduced, and robust phase correction can be performed.

[0095] [Step S806]

[0096] The x-direction inclination calculation unit 513 calculates the inclination in the x direction using the differential phase c m (x, z) obtained in Step S805. Therefore, first, the x-direction phase differentiation calculation unit performs complex accumulation through Equation (7-1), and obtains the phase differentiation (phase change per pixel) c d in the x direction through Equation (7-2). Next, the x-direction average calculation unit performs a complex average on the phase differentiation c d of Equation (7-2) using Equation (8), and uses it as the inclination α φ in the x direction.

[0097]

Mathematical formula 7

[0098]

[0099]

[0100] In Equations (7-1) and (7-2), the subscript characters "ds" and "d" of "c" represent complex accumulation and phase differentiation, respectively.

[0101]

Mathematical formula 8

[0102]

[0103] In Equation (8), nx represents the number of pixels in the x direction (the same applies hereinafter).

[0104] In Equations (7-1) and (7-2), by using complex accumulation in the calculation of phase differentiation, the signal value (|c ds |) is included as the weight in the complex averaging in Equation (8), improving robustness.

[0105] [Step S807]

[0106] The x-direction intercept calculation unit 514 uses the inclination α of the phase calculated in Step S806 φ to calculate the intercept. Therefore, first, the phase at the reference position is obtained. The reference position is arbitrary. For example, it is set as the Fourier center (the central position of the data). The phase (temporary phase) φ0 in this case can be represented by Equation (9). In Equation (9), x0 is the coordinate of the reference position (Fourier center).

[0107]

Mathematical Equation 9

[0108] φ0(x, z) = α φ (z)(x - x0) (9)

[0109] Next, from the temporary phase φ0, as its difference, the intercept s is obtained through Equation (10), and its complex average φ s is calculated through Equation (11), and it is used as the correction phase. Through such processing, the correction phase without deviation of the intercept can be obtained.

[0110]

Mathematical Equation 10

[0111] s(x, z) = c m (x, z)·exp{-i·φ0(x, z)} (10)

[0112]

Mathematical Equation 11

[0113]

[0114] [Step S808]

[0115] The phase correction unit 515 uses the temporary phase φ0 and the complex average φ of the intercept obtained in Step S807 s to calculate the correction phase (phase correction value) through Equation (12), and corrects the phase in the x direction of the formal scan data through Equation (13).

[0116]

Mathematical Equation 12

[0117] φ fit (x, z) = φ0(x, z) + φ s (z) (12)

[0118]

Mathematical formula 13

[0119] h(x, k y , z) = exp(+i·φfit(x, z) / 2) (13 - 1)

[0120] h(x, k y , z) = exp(-i·φfit(x, z) / 2) (13 - 2)

[0121] In Figure 11 shows an example of the phase for correction. In Figure 11 , the phases shown by dotted lines respectively correspond to Equation (13 - 1) and Equation (13 - 2). When performing corrections on the odd - numbered rows and even - numbered rows of the formal scan data respectively, use them for phase correction as follows.

[0122] For the data of the odd - numbered rows in the first formal scan data and the data of the even - numbered rows in the second formal scan data, perform complex division operation on h(x, ky, z) from the original x - ky space data using Equation (13 - 1). In addition, for the data of the even - numbered rows in the first formal scan data and the data of the odd - numbered rows in the second formal scan data, perform complex division operation on h(x, ky, z) from the original x - ky space data using Equation (13 - 2).

[0123] In addition, it is also possible to use Figure 11 the phase for correction (2 times the value) shown by the solid line in

[0124] to perform correction only on the even - numbered rows and make it consistent with the odd - numbered rows.

[0125] Thus, the x - ky space data with x - direction phase correction is obtained. Figure 11 The above is the phase correction performed by the x - direction correction unit 51. In this way, in the x - direction phase correction, since the phase correction value, that is, the inclination and intercept in the x - direction, are all calculated by complex averaging, a phase correction value that maintains the magnitude of the signal value can be obtained. In addition, without performing unstable phase unwrapping and fitting processing, the phase for correction as shown in

[0126] Next, refer to Figure 12 to detail the processing of the ky - direction correction unit 52. Figure 12 is the flow chart showing the processing in the ky - direction.

[0127] Correction in the ky direction processes correction data (x-ky space data created from k-space data obtained through pre-scanning) by dividing it into multiple patterns according to the excitation number and the parity of the rows. Therefore, first, the correction data is segmented (S901). Next, after performing phase differentiation processing in the ky direction (S902) and complex averaging processing on each of the segmented patterns to calculate the inclination in the ky direction for each slice (S903), the inclinations in the ky direction calculated for each slice of each pattern are averaged to calculate the correction data (S904). Finally, the phase correction of the formal scan data is performed using the correction data (S905).

[0128] The following details each step of the ky direction correction will be described.

[0129] [Step S901]

[0130] In this embodiment, since the pre-scan data is obtained by imaging twice for each channel, the x-ky space data is divided according to each imaging (the first correction data, the second correction data), and they are further divided into data of odd rows and data of even rows. Thus, the x-ky space data is divided into four patterns: the first odd row, the first even row, the second odd row, and the second even row.

[0131] For the data p(Ky, x, z, ch, nsa) (nsa is the number of each imaging, which can take 1 or 2 in this embodiment) before segmentation, the data after segmentation is described as p 1o (Ky’, x, z, ch), p 1e (Ky’, x, z, ch), p 2o (Ky’, x, z, ch), p 2e (Ky’, x, z, ch). Here, in order to distinguish the ky direction coordinates before and after segmentation, the former is set as Ky, and the latter is set as Ky’.

[0132] [Step S902]

[0133] Next, the phase differentiation (the inclination in the Ky’ direction) in the Ky’ direction is calculated for the data of each pattern. Similar to the case of obtaining the inclination in the x direction (Equations (7-1) and (7-2)), the phase differentiation is obtained through the complex accumulation between adjacent rows, that is, by accumulating with one set as the complex conjugate. As a representative, the calculation formulas for the data p 1o (Ky’, x, z, ch) of one pattern are shown as the following equations (14-1) and (14-2). The same calculation is performed for other patterns to obtain the phase differentiation.

[0134]

Mathematical formula 14

[0135]

[0136]

[0137] In Expressions (14-1) and (14-2), the subscript characters "ds" and "d" of "p 1o " respectively represent complex accumulation and phase differentiation.

[0138] Here too, by using complex accumulation in the calculation of phase differentiation, the signal value (|p ds |) is included as a coefficient of phase differentiation, that is, a weight, to enhance the robustness of the result. Through this Step S902, a correction amount corresponding to the position deviation in the y direction is obtained.

[0139] [Step S903]

[0140] The result obtained in Step S902 is subjected to complex averaging between pixels in the Ky' direction and the x direction, and between channels (Expression (15)).

[0141]

Mathematical Expression 15

[0142]

[0143] Thus, the position deviation in the y direction (the inclination in the ky direction) is obtained for each pattern. In Figure 13 An example of the change in data of the processing based on the above Steps S901 to S903 is shown. Figure 13 (A) of shows the phase diagrams of the first and second x-ky spatial data of the central slice (for example, the 10th slice) of the imaging range, and (B) shows the phase diagram after being divided into patterns. By performing processing in such a divided pattern, as shown in (C), the results of the phase differentiation are substantially the same in each pattern.

[0144] In this way, by dividing the data according to odd and even rows for processing, without assuming x-direction phase correction, the phase correction value in the ky direction can be calculated. Furthermore, by performing processing in a divided pattern, the deviation caused by the pattern can be averaged, and the accuracy of the phase correction value can be improved.

[0145] In addition, in Expression (15), the average is calculated for Ky', x, and channels, but the average can also be calculated only for Ky' and channels. In this case, ps becomes a function ps(x, z) related to x and z. If this ps(x, z) is used for the subsequent phase correction processing, as in the present embodiment, not only the difference in the inclination in the ky direction for each slice can be corrected, but also the difference in the inclination in the ky direction for each position x can be corrected.

[0146] [Step S904]

[0147] In step S903, the average of the inclinations in the ky direction calculated by formula (15) for the four patterns is taken to calculate the correction value. Since the inclination in the ky direction of each pattern is calculated based on the data skipping one row (odd / even respectively), when taking the average, the sum of the angles of the inclinations is divided not by the number of patterns 4 but by 2 as shown in the following formula (16).

[0148]

Mathematical formula 16

[0149]

[0150] As a result, as Figure 14 shown, the inclination φ in the ky direction is obtained s .

[0151] Next, using this inclination φ s the correction value is calculated by the following formula (17).

[0152]

Mathematical formula 17

[0153] h s (x, k y , z) = exp(i·k y ·φ s (z)) (17)

[0154] In addition, in the imaging by the PI method, when the matrix size of the k-space data for the formal scan data and the matrix size of the pre-scan are different, etc., a coefficient for adjusting the size can be multiplied by φ in formula (17) corresponding to the magnification rate. s

[0155] [Step S905]

[0156] The formal scan data that has undergone the aforementioned phase correction in the x direction is corrected with the correction value calculated in step S904. This correction is performed in the x-ky space. That is, the formal scan data (k-space data) is Fourier-transformed in the x direction to obtain x-ky space data, and the correction coefficient hs is cumulatively complexed for it. Thus, a correction equivalent to the correction of shifting in the y direction in the real space is performed.

[0157] Through the above steps, the phase correction in the ky direction is completed.

[0158] ​In the above-described phase correction in the ky direction, the correction data is divided into four patterns to calculate the phase correction value, and the results are combined as the phase correction value. Therefore, the calculation accuracy of the inclination is improved. In addition, since the odd-numbered rows and the even-numbered rows are divided and processed, it is not premised on the phase correction in the x direction. Therefore, it can be performed as a process independent of the phase correction process in the x direction. Furthermore, by combining the four patterns by complex averaging, a phase correction value reflecting the absolute value can be obtained, and phase unwrapping and fitting processes are not required, and the inclination can be calculated stably.

[0159] Finally, the image reconstruction unit 24 reconstructs an image from the officially scanned data after phase correction. At this time, as described above, since the correction in the x direction and the correction in the y direction can be performed independently, the phase correction of the officially scanned data can be performed in either order.

[0160] As described above, according to the present embodiment, the magnitude (absolute value) of the signals different for each channel is used as a weight, and the phase correction values obtained from the correction data obtained for each channel are combined as a common phase correction value. Thereby, the aliasing artifacts that may occur when phase correction is performed for each channel can be suppressed, and the phase correction of the measurement data obtained by EPI can be performed with good accuracy.

[0161] In addition, according to the present embodiment, since complex averaging is used when combining the correction values in the x direction and the correction values in the ky direction between channels, the phase unwrapping process and the fitting process that are generally likely to become unstable can be avoided, and the phase correction value can be calculated stably.

[0162] In Figure 15 (A) and (B) show the image (A) reconstructed by performing phase correction in the present embodiment, and the image (B) reconstructed by calculating the phase correction value for each channel and performing PI synthesis after phase correction.

[0163] As can be seen by comparing these figures, when phase correction is performed for each channel, as shown by the black arrow in Figure 15 (B), aliasing artifacts occur in a part of the image, but by performing the process of the present embodiment, the generation of such artifacts is suppressed.

[0164] <Modification Example 1 of the Embodiment>

[0165] In Embodiment 1, the case where the imaging unit 10 performs imaging in the sequence of the PI method has been described. However, the present invention can also be applied to imaging at a double rate, that is, imaging without undersampling, as long as it is EPI using a multi-channel receiving coil.

[0166] In addition, in Embodiment 1, pre-scan data is used to calculate the correction phase, but formal scan data can also be used to calculate the correction phase. Among them, since the formal scan data is obtained by applying an inclined magnetic field for phase encoding, it is difficult to independently calculate the phase error in the ky direction caused by magnetic field inhomogeneity or the like. Therefore, in the case of performing phase correction using formal scan data, only the phase correction in the x direction is performed. In this case, although the phase correction in the ky direction cannot be performed, the imaging time can be shortened because pre-scanning is not required.

[0167] In addition, in Embodiment 1, the case of calculating the phase correction value using the data of two single excitations in opposite directions of scanning the k-space is described, but as long as it is a scan that can obtain a group of row data (odd row data and even row data) in opposite scanning directions, the data of one time can also be used, and the data of multiple excitations can also be used.

[0168] In the case of one time, the odd row and the subsequent even row (for example, the row with ky = 0 and its adjacent row) are taken as one group to perform the above-mentioned x-direction correction. In addition, in the ky-direction correction, after calculating the phase correction value by dividing into two patterns of odd rows and even rows, they can be synthesized.

[0169] In addition, the present invention creates and applies common correction data for phase-correcting multi-channel data by synthesizing the correction data of each channel, and can also be applied in the case where only one of the phase corrections in the X direction or the ky direction is implemented. Such a technique is also included in the present invention.

[0170] The application of the present invention to an MRI apparatus has been described above, but the present invention can also be applied to an image processing apparatus 2 independent of the MRI apparatus. Figure 16 It is a diagram showing an embodiment of an image processing apparatus to which the present invention is applied.

[0171] As shown in the figure, this image processing apparatus mainly includes: a computer 20; a UI unit 30 including a display device, an input device, etc. attached thereto; and a storage device 40. The computer 20 is connected to one or more MRI apparatuses 1 via a network or the like, and receives the data imaged by the MRI apparatus via these networks. Alternatively, it can also be received via a removable medium or the like.

[0172] The structure of the computer 20 is the same as Figure 4 and Figure 6The structures of the computer and the preprocessing unit shown are the same, and they perform the same operations. That is, using the correction data collected by the MRI device for each of the multiple channels, the phase correction values are calculated for the x direction and the ky direction, and they are synthesized to calculate a common phase correction value, and this phase correction value is applied to the image data for phase correction. The specific processing content is also the same, and repeated explanations are omitted.

Claims

1. A magnetic resonance imaging apparatus, characterized in that, Comprising: An imaging unit that applies high-frequency magnetic field pulses and gradient magnetic field pulses in accordance with a given pulse sequence, and collects nuclear magnetic resonance signals generated from a subject; and An arithmetic unit that uses the nuclear magnetic resonance signals to generate an image of the subject, The imaging unit includes a multi-channel receive coil, and as the pulse sequence, executes a pulse sequence of an EPI method that acquires a plurality of nuclear magnetic resonance signals after applying an excitation high-frequency magnetic field pulse once, The arithmetic unit includes a preprocessing unit that corrects the phase of data composed of nuclear magnetic resonance signals respectively obtained in a plurality of channels of the receive coil by executing the pulse sequence of the EPI method, The preprocessing unit comprises: A correction value calculation unit that calculates a phase correction value using correction data respectively received in a plurality of channels, and synthesizes the phase correction values of each channel to calculate a common phase correction value; And A phase correction unit that corrects the phase of image formation data received in each channel using the common phase correction value, The correction value calculation unit calculates the phase correction value of each channel as a differential phase based on the complex accumulation between a set of row data, and calculates the common phase correction value by the complex average of the differential phases.

2. The magnetic resonance imaging apparatus according to claim 1, wherein The correction value calculation unit calculates and synthesizes the differential phases calculated for each channel for a plurality of slices.

3. The magnetic resonance imaging apparatus according to claim 2, wherein The correction value calculation unit uses a Gaussian kernel function in the synthesis of the differential phases of a plurality of slices.

4. The magnetic resonance imaging apparatus according to claim 1, wherein The preprocessing unit transforms the correction data into x-ky space data, and uses the x-ky space data to calculate the phase correction value and perform phase correction.

5. The magnetic resonance imaging apparatus according to claim 4, wherein The preprocessing unit includes an x-direction correction unit that performs phase correction in the x direction and a ky-direction correction unit that performs phase correction in the ky direction, The common phase correction value is calculated respectively for the x-direction correction unit and the ky-direction correction unit.

6. The magnetic resonance imaging apparatus according to claim 5, wherein The x-direction correction unit calculates a differential phase for each channel by complex accumulation between a set of row data in the x-ky space, calculates an average differential phase between channels by complex averaging the differential phases of each channel, and calculates an inclination in the x direction and an x-intercept as the common phase correction value in the x direction.

7. The magnetic resonance imaging apparatus according to claim 5, wherein The ky-direction correction unit includes a data segmentation unit that segments the correction data of each channel into a plurality of patterns corresponding to the number of excitations of the EPI and the parity of the row data, After calculating the phase correction value for each pattern segmented by the data segmentation unit, the phase correction values of each pattern are synthesized as the common phase correction value in the ky direction.

8. The magnetic resonance imaging apparatus according to claim 1, characterized in that the data for correction is data obtained by the imaging unit through pre-scanning.

9. The magnetic resonance imaging apparatus according to claim 8, characterized in that the imaging unit uses the same EPI sequence to acquire the data for correction and the data for image formation.

10. The magnetic resonance imaging apparatus according to claim 9, characterized in that the imaging unit executes two EPI sequences with opposite scan directions in the k-space, and the pre-processing unit calculates a phase correction value by using the data obtained by the two EPI sequences as a set of data.

11. The magnetic resonance imaging apparatus according to claim 1, characterized in that the operation unit includes an image reconstruction unit that reconstructs the k-space data obtained by the imaging unit through undersampling by parallel imaging operation.

12. An image processing apparatus, wherein a magnetic resonance imaging apparatus processes data received by each channel of a multi-channel receiving coil to reconstruct an image, and the image processing apparatus is characterized by comprising: a correction value calculation unit that calculates a phase correction value by using the correction data received in the plurality of channels respectively, and synthesizes the phase correction values of each channel to calculate a common phase correction value; a phase correction unit that corrects the phase of the image formation data received by each channel by using the common phase correction value; and a reconstruction unit that reconstructs an image by using the image formation data whose phase has been corrected, the correction value calculation unit calculates the phase correction value of each channel as a differential phase based on the complex accumulation between a set of row data, and calculates the common phase correction value by the complex average of the differential phases.

13. A phase correction method for correcting the phase of image formation data respectively acquired in a plurality of channels by an EPI sequence in a magnetic resonance imaging apparatus having a multi-channel receiving coil, the phase correction method being characterized by including: a step of calculating a phase correction value by using the correction data received in the plurality of channels respectively; a step of synthesizing the phase correction values of each channel to calculate a common phase correction value; and a step of correcting the phase of the image formation data received by each channel by using the common phase correction value, the step of calculating the phase correction value includes a step of calculating the differential phase between rows in the x-ky space data of the correction data by complex accumulation, and calculating the phase correction value based on the differential phase, the step of calculating the common phase correction value includes a step of performing cumulative averaging on the phase correction values of each channel.

14. The phase correction method according to claim 13, characterized in that the step of calculating the phase correction value includes a step of calculating the phase correction value in the x direction of the x-ky space and a step of calculating the phase correction value in the ky direction, the step of calculating the phase correction value in the x direction and the step of calculating the phase correction value in the ky direction are independently executed.

Images