Image processing method and image processing apparatus

Abstract

To give proper motion correction to MRI data.SOLUTION: Receiving k-space data acquired by scanning an object with a magnetic resonance imaging apparatus, wherein the k-space data includes a first set of motion corrupted k-space data and a second set of k-space data different from the first set of motion corrupted k-space data, generating motion corrected data based on the first set of motion corrupted k-space data and information indicating whether the object moved when scanning the object, and generating an image based on the second set of k-space data and the motion corrected data.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] FIELD Embodiments of the present invention relate to an image processing method and an image processing device. [Background technology]

[0002] Motion artifacts are a common problem in medical imaging, such as magnetic resonance imaging (MRI), due at least in part to long acquisition times. As described in U.S. Patent Application Publication No. 2007 / 0129994, images with motion can be detected and rejected.

[0003] The navigator corrects for in-plane rigid body motion and can also be used for shot rejection. For example, (1)Lin, W., Huang, F., B▲o▼rnert, P., Li, Y. and Reykowski, A., 2010; Motion correction using an enhanced floating navigator and GRAPPA operations; Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine, 63(2), pp. 339-348;(2)Kober, T., Marques, JP, Gruetter, R. and Krueger, G., 2011. Head motion detection using FID navigators. Magnetic resonance in medicine, 66(1), pp. 135-143; and (3) Wallace, TE, Afacan, O., Waszak, M., Kober, T. and Warfield, SK, 2019. Head motion measurement and correction using FID navigators. Magnetic resonance in medicine, 81(1), pp. 258-274, the contents of each of these three references are incorporated herein by reference.

[0004] Post-processing iterative methods can determine unknown patient motion and correct for motion artifacts using a focus criterion related to entropy. See, e.g., Atkinson, D., Hill, DL, Stoyle, PN, Summers, PE, Clare, S., Bowtell, R. and Keevil, SF, 1999. Automatic compensation of motion artifacts in MRI. Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine, 41(1), pp. 163-170, the contents of which are incorporated herein by reference.

[0005] Aligned SENSE jointly estimates rigid body motion parameters and reconstructs MR images using an iterative reconstruction framework that relies on redundant information provided by multiple coils, as described in Cordero-Grande, L., Teixeira, R.P.A., Hughes, E.J., Hutter, J., Price, A.N. and Hajnal, J.V., 2016. Sensitivity encoding for aligned multishot magnetic resonance reconstruction. IEEE Transactions on Computational Imaging, 2(3), pp. 266-280, the contents of which are incorporated herein by reference.

[0006] A promising camera-based approach allows for real-time motion tracking and dynamic update of acquisition geometry, as described by Siemens partnering with KinetiCor on a four-camera in-bore system: https: / / www.siemens-healthineers.com / press-room / press-releases / pr-20180617020shs.html.

[0007] In one known method of motion compensation by shot rejection, motion-corrupted portions of k-space (called "shots") are removed from the k-space data, and the resulting undersampled k-space is used for image reconstruction. However, such shot-rejection-based techniques can lead to degradation of image quality, as will be described below.

[0008] As shown in FIG. 1A, k-space data is a set of imaging data at various frequencies in the frequency domain, which can be transformed into an image in the spatial domain. The frequency domain information is contained in the central portion of k-space (e.g., f 1,1 ,f -1,1 ,f 1,-1 ,f -1,-1 ) and the frequency is low and increases as it moves away from the center. n,n ,f n,-n ,f -n,n , and f -n,-n) where the frequency is highest. To generate some types of images from k-space data, not all frequencies (also called lines) need to be acquired (sampled); a data set that is not fully sampled is called an undersampled data set. The low-frequency central portion of the k-space data may be fully sampled during the scan (shown within the bold box in the center of the k-space data in FIGS. 1A and 1B ) so that the frequency is used as an auto-calibration signal (ACS).

[0009] Additionally, for data removed from the k-space data set due to intentional undersampling, such as patient movement during imaging, an indication of motion may be stored as part of the imaging or referenced in conjunction with accessing the imaging data. Thus, for each k-space point or line, the representative data may include whether the data is present (P), whether the data was lost due to undersampling (X), and, if present, the magnitude and phase of each sampled point within the set of points comprising the k-space line, and whether motion (M) was present. To aid in describing the invention herein, the magnitude and phase of each sampled k-space point may be referred to using the letter (S), and the notation (P, X, S, M) may be added to a box representing the acquired k-space imaging data. For example, in FIG. 1B, all of the low frequency data is present and includes sampled magnitude and phase values ​​(S4, S5, S12, S13), but the data at frequency f x,1k-space data was acquired in the presence of motion. To avoid clutter, blank boxes contain k-space data and are considered motionless. The magnitude and phase (S) of the sampled points may also be omitted from the plot of the current motionless data. Typically, a full readout will have similar behavior (e.g., motion corrupted or undersampled). Readouts can take different forms. One such form is a line in Cartesian space. In this scenario, all k-space points along the line will have identical behavior.

[0010] As shown in FIG. 2, in known systems, motion-corrupted k-space data is received by a system for generating a spatial image (e.g., a medical image). In the illustrated example, there are two lines known to have been acquired in the presence of motion (M), and the system may choose to remove these lines when generating the spatial image. As shown in FIG. 2, the lines corresponding to sampled values ​​S5 through S9 were both acquired in the presence of motion (as noted in the data itself or detected via navigator signals and / or other motion detection circuitry). As a result of performing motion shot rejection on the motion-corrupted k-space data, two lines are marked as missing in k-space (marked with an X) and will be filled in with approximate data in the undersampled reconstruction. Additionally, the undersampled k-space may have additional missing lines due to undersampling or errors.

[0011] Known undersampled reconstruction methods, such as Compressed Sensing (CS) reconstruction and Deep Learning Reconstruction (DLR), utilize a complete central ACS during reconstruction. For example, as shown in FIG. 2, ACS data can be used by a sensitivity map generator to generate a sensitivity map. However, in the illustrated embodiment, the ACS used to help determine how to fill in the missing lines itself has motion within it, which would reduce the ability of the sensitivity map to aid in undersampled reconstruction. As a result, the resulting spatial image may be of low quality, even though it is typically motion-compensated.

[0012] An external ACS can be acquired separately from the imaging data, reducing the ACS sensitivity to motion. External ACS is limited by (a) the potential for mismatch between the ACS and the imaging data due to motion, (b) the potential for ACS corruption due to motion, (c) the potential for reduced reconstruction performance when the external ACS and the imaging data are acquired with different sequence parameters, and / or (d) the increased scan time required for external ACS. [Prior art documents] [Patent documents]

[0013] [Patent Document 1] U.S. Patent No. 9,710,937 [Patent Document 2] US Patent Application Publication No. 20220187406 Summary of the Invention [Problem to be solved by the invention]

[0014] One of the problems to be solved by the embodiments disclosed in this specification and the drawings is to provide appropriate motion correction to MRI data. However, the problems to be solved by the embodiments disclosed in this specification and the drawings are not limited to the above problem. Problems corresponding to the effects of the configurations shown in the embodiments described below can also be positioned as other problems. [Means for solving the problem]

[0015] An image processing method of an embodiment includes receiving k-space data acquired by scanning an object with a magnetic resonance imaging device, the k-space data including a first set of motion-corrupted k-space data and a second set of k-space data different from the first set of motion-corrupted k-space data; generating motion-corrected data based on the first set of motion-corrupted k-space data and information indicating whether the object moved while scanning the object; and generating an image based on the second set of k-space data and the motion-corrected data. [Brief explanation of the drawings]

[0016] [Figure 1A] FIG. 1A is a schematic diagram of an exemplary frequency / line of k-space imaging data with fully sampled central k-space data serving as an auto-calibration signal (ACS). [Figure 1B] FIG. 1B is a schematic diagram of exemplary k-space imaging data showing whether various lines are present, whether motion was present when the corresponding line was sampled, and what the energy data for each line is. [Figure 2] FIG. 2 is a block diagram illustrating a known image generation architecture that utilizes undersampled reconstruction based on corrupted / uncorrupted ACS, resulting in a degradation of image quality. [Figure 3] FIG. 3 is a schematic diagram of an MRI apparatus. [Figure 4]FIG. 4 is a flowchart illustrating a generalized process described herein for generating an image from k-space data, the k-space data including a first set of motion-corrupted k-space data and a second set of undersampled k-space data, by (a) generating motion-compensated data based on the first set of motion-corrupted k-space data, and (b) generating an image based on the second set of undersampled k-space data and the motion-compensated data. [Figure 5A] FIG. 5A is a block diagram illustrating an image generation architecture that utilizes motion-compensated ACS-based undersampled reconstruction, resulting in improved image quality. [Figure 5B] FIG. 5B is a block diagram illustrating the image generation architecture of FIG. 5A utilizing compressed sensing (CS) undersampled reconstruction using sensitivity maps as correction data. [Figure 5C] FIG. 5C is a block diagram illustrating the image generation architecture of FIG. 5A utilizing GRAPPA undersampled reconstruction using GRAPPA weights as correction data. [Figure 5D] FIG. 5D is a block diagram illustrating the image generation architecture of FIG. 5A utilizing deep learning undersampled reconstruction (DLR) using sensitivity maps (e.g., ESPIriT) as correction data. [Figure 6A] FIG. 6A is a plot of the navigator's correlation information from a series of shots. [Figure 6B] FIG. 6B is an annotated plot of the navigator's correlation information from the sequence of shots in FIG. 6A, annotated to indicate shots that were rejected due to low correlation values. [Figure 7A] FIG. 7A is an additionally annotated plot of the navigator correlation information corresponding to the images of FIGS. 7B and 7D, annotated to indicate shots that were rejected due to low correlation values. [Figure 7B]FIG. 7B is an ACS (sum of squares) image corresponding to the correlation information for the scan of FIG. 7A, which includes motion corruption in the ACS information. [Figure 7C] FIG. 7C is an ACS motion corrected version of FIG. 7B. [Figure 7D] FIG. 7D is a graphical representation of ACS k-space data from an ACS k-space coil but containing motion corruption in the ACS information. [Figure 7E] FIG. 7E is an ACS motion corrected version of FIG. 7D. [Figure 8A] FIG. 8A is a first block diagram of the interior of the motion compensation circuit according to the first embodiment at three different iterations of the GRAPPA reconstruction. [Figure 8B] FIG. 8B is a second block diagram of the interior of the motion compensation circuit according to the first embodiment at three different iterations of the GRAPPA reconstruction. [Figure 8C] FIG. 8C is a third block diagram of the interior of the motion compensation circuit according to the first embodiment at three different iterations of the GRAPPA reconstruction. [Figure 9] FIG. 9 is a block diagram of a motion compensation circuit according to a second embodiment using multi-step GRAPPA / RAKI processing. [Figure 10] FIG. 10 is a block diagram illustrating an alternative image generation architecture that utilizes motion-compensated ACS-based undersampled reconstruction. [Figure 11A] Figure 11A is an image produced using undersampled compressed sensing (CS) reconstruction on a set of motion-corrupted k-space data. [Figure 11B] FIG. 11B is the image that results from performing motion correction on the motion-corrupted k-space data set corresponding to FIG. 11A before performing undersampled CS reconstruction. [Figure 11C]FIG. 11C is an image resulting from performing another motion correction on the ACS k-space data corresponding to FIG. 11A before performing undersampled CS reconstruction on the motion-corrected combined k-space data and before performing motion correction on the motion-corrected combined ACS k-space data and non-ACS k-space data. [Figure 12A] FIG. 12A is an image produced using GRAPPA reconstruction on a motion-corrupted k-space data set. [Figure 12B] FIG. 12B is the image that results from performing motion correction (shot rejection) on the motion-corrupted k-space data set corresponding to FIG. 12A before performing GRAPPA reconstruction. [Figure 12C] FIG. 12C is an image resulting from performing another motion correction on the ACS k-space data corresponding to FIG. 12A before performing GRAPPA reconstruction on the motion-corrected combined k-space data and before performing motion correction on the motion-corrected combined ACS k-space data and non-ACS k-space data. DETAILED DESCRIPTION OF THE INVENTION

[0017] The term "plurality," as used in embodiments, is defined as two or more than two. The term "another," as used herein, is defined as at least a second or more. The terms "including" and / or "having," as used herein, are defined as comprising (i.e., open language). References throughout this document to "one embodiment," "certain embodiment," "an embodiment," "an implementation," "an example," or similar terms mean that the specific feature, structure, or characteristics described in connection with that embodiment are included in at least one embodiment of the present disclosure. As such, the appearances of such phrases or phrases in various places throughout this specification need not all refer to the same embodiment. Furthermore, specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments without limitation.

[0018] The present disclosure relates to methods, systems, and non-transitory computer-readable storage media storing computer-readable instructions for providing initial motion correction in imaging data (e.g., magnetic resonance imaging (MRI) data) that can provide additional image correction to subsequently processed imaging data within the same data set.

[0019] In one embodiment, the present disclosure can be preferably viewed as a system, and although the present exemplary embodiment refers to an MRI device, other system configurations can preferably use other medical imaging devices (e.g., CT systems and combined MRI / CT systems).

[0020] Now, referring to the drawings, Fig. 3 is a block diagram showing a schematic configuration of an MRI apparatus 1. The MRI apparatus 1 includes a gantry 100, a control cabinet 300, a console 40, a bed 50, and a radio frequency (RF) coil 20. The gantry 100, the control cabinet 300, and the bed 50 constitute a scanner, i.e., an imaging unit.

[0021] The gantry 100 includes a static magnetic field magnet 10, a gradient magnetic field coil 11, and a whole body (WB) coil 12, and these components are housed in a cylindrical housing. The bed 50 includes a bed body 52 and a base 51.

[0022] The control cabinet 300 includes three gradient coil power supplies 31 (31x for the X axis, 31y for the Y axis, and 31z for the Z axis), a coil selection circuit 36, an RF receiver 32, an RF transmitter 33, and a sequence controller 34.

[0023] The console 40 includes a processing circuit 45, a memory 41, a display 42, and an input interface 43. The console 40 functions as a host computer.

[0024] The static magnetic field magnet 10 of the gantry 100 is a generally cylindrical body that generates a static magnetic field within a bore through which an object, such as a patient, is carried. The bore is the internal space of the cylindrical structure of the gantry 100. The static magnetic field magnet 10 contains a superconducting coil inside, which is cooled to a very low temperature by liquid helium. The static magnetic field magnet 10 generates a static magnetic field by supplying a current given by a static magnetic field power supply (not shown) to the superconducting coil in an excitation mode. The static magnetic field magnet 10 then transitions to a persistent current mode, and the static magnetic field power supply is disconnected. Once in the persistent current mode, the static magnetic field magnet 10 continues to generate a strong static magnetic field for a long period of time, such as one year. In Figure 3, the black circle on the chest of the object indicates the magnetic field center.

[0025] The gradient magnetic field coil 11 is also a substantially cylindrical body, and is fixed inside the static magnetic field magnet 10. The gradient magnetic field coil 11 applies gradient magnetic fields (e.g., gradient pulses) to the object in the directions of the X-axis, Y-axis, and Z-axis using currents supplied from gradient magnetic field coil power supplies 31x, 31y, and 31z.

[0026] The bed body 52 of the bed 50 can move the table 51 in the vertical and horizontal directions. Before imaging, the bed body 52 moves the table 51, on which an object is placed, to a predetermined height. Then, when imaging the object, the bed body 52 moves the table 51 in the horizontal direction to move the object into the bore.

[0027] The WB coil 12 is a substantially cylindrical body that surrounds the object and is fixed inside the gradient magnetic field coil 11. The WB coil 12 applies RF pulses transmitted from an RF transmitter 33 to the object. The WB coil 12 also receives magnetic resonance signals, or MR signals, emitted from the object due to excitation of hydrogen nuclei.

[0028] In addition to the WB coil 12, the MRI apparatus 1 may include RF coils 20 shown in FIG. 3. Each RF coil 20 is a coil placed near the body surface of the object. There are various types of RF coils 20. For example, as shown in FIG. 3, types of RF coils 20 include a body coil attached to the chest, abdomen, or legs of the object, and a spine coil attached to the back of the object. Another type of RF coil 20 is a head coil for imaging the head of the object. Most RF coils 20 are receive-only coils, but some RF coils 20, such as head coils, are of a type that perform both transmission and reception. The RF coil 20 is configured to be attachable to and detachable from the base 51 via a cable.

[0029] The RF transmitter 33 generates each RF pulse based on an instruction from the sequence controller 34. The generated RF pulse is transmitted to the WB coil 12 and applied to the object. MR signals are generated from the object by the application of one or more RF pulses. Each MR signal is received by the RF coil 20 or the WB coil 12.

[0030] The MR signals received by the RF coil 20 are transmitted to the coil selection circuit 36 ​​via cables in the table 51 and the bed body 52. ​​The MR signals received by the WB coil 12 are also transmitted to the coil selection circuit 36.

[0031] The coil selection circuit 36 ​​selects the MR signals output from each RF coil 20 or the MR signals output from the WB coil in response to a control signal output from the sequence controller 34 or the console 40.

[0032] The selected MR signals are output to the RF receiver 32. The RF receiver 32 performs analog-to-digital (AD) conversion on the MR signals and outputs the converted signals to a sequence controller 34. In some cases, the digitized MR signals are referred to as raw data. The AD conversion may be performed inside each RF coil 20 or inside the coil selection circuit 36.

[0033] The sequence controller 34 scans an object by driving the gradient magnetic field coil power supply 31, the RF transmitter 33, and the RF receiver 32 under the control of the console 40. By performing the scan, the sequence controller 34 receives raw data from the RF receiver 32, and transmits the received raw data to the console 40.

[0034] The sequence controller 34 includes a processing circuit (not shown). The processing circuit is configured as, for example, a processor for executing a predetermined program, or as hardware such as a field programmable gate array (FPGA) or an application specific integrated circuit (ASIC).

[0035] As described above, the console 40 comprises a memory 41, a display 42, an input interface 43, and a processing circuit 45.

[0036] The memory 41 is a storage medium that includes a read-only memory (ROM) and a random access memory (RAM) in addition to an external storage device such as a hard disk drive (HDD) or an optical disk device. The memory 41 stores various programs executed by the processor of the processing circuitry 45, as well as various data and information.

[0037] The input interface 43 includes various devices for the operator to input various information and data, and is configured, for example, by a mouse, a keyboard, a trackball, and / or a touch panel.

[0038] The display 42 is a display device such as a liquid crystal display panel, a plasma display panel, or an organic EL panel.

[0039] The processing circuit 45 is a circuit including, for example, a central processing unit (CPU) and / or a special-purpose or general-purpose processor. The processor executes programs stored in the memory 41 to realize various functions (e.g., method 400) described below. The processing circuit 45 may be configured as hardware such as an FPGA or an ASIC. The various functions described below can be realized by such hardware. In addition, the processing circuit 45 can realize various functions by combining hardware processing and software processing based on the processor and programs.

[0040] FIG. 4 is a flowchart illustrating a general process described herein. In method 400, processing begins at step 410 with receiving k-space data including a first set of motion-corrupted k-space data and a second set of k-space data different from the first set. Each of the first and second sets of k-space data may be undersampled k-space data, parallel-sampled data, or fully-sampled k-space data. At step 420, motion-corrected data is generated based on the first set of motion-corrupted k-space data. The motion-corrected data may be (1) information (e.g., GRAPPA weights or sensitivity maps such as ESPIriT) to aid in (1a) generating interpolated data to "fill in" missing k-space data and / or (1b) reconstruction after undersampling, (2) new k-space data to replace all or part of the first set of motion-corrupted k-space data, or (3) a combination of (1) and (2). In step 430, a spatial image is generated based on the second set of k-space data and the motion-corrected data (e.g., using undersampled or full reconstruction). In embodiments where the second set of k-space data is undersampled k-space data, the method may "fill in" the missing k-space data using interpolation to generate interpolated k-space data from the second set of k-space data (and potentially the motion-corrected data (e.g., sensitivity maps)). In such embodiments, the spatial image is generated based on the second set of k-space data, the interpolated k-space data (e.g., generated using the sensitivity maps), and the motion-corrected data.

[0041] FIG. 5A is a block diagram illustrating a generalized image generation architecture for implementing the method of FIG. 4. System 500 (and its associated methods and / or computer program products) utilizes undersampled reconstruction 580 based on a first set of motion-corrected k-space data used to correct a second set of undersampled k-space data to improve image quality. Alternatively, if the second set of k-space data is a fully sampled k-space, the sampled data can be reconstructed. As shown in FIG. 5A, an initial set of k-space data 505 is received within system 500 (as in step 410 of FIG. 4), which includes a first set of motion-corrupted k-space data 507 and a second set of k-space data outside the first set. The first set of motion-corrupted k-space data 507 is shown to include the central lines / frequencies along with k-space points having sampled magnitudes and phases S4, S5, S12, and S13 as shown, although this first set can include additional lines (e.g., all lines of the ACS, a majority of the lines of the ACS, or substantially all (e.g., 90%) of the lines of the ACS). The k-space points having sampled magnitudes and phases S4 and S5 are indicated (labeled "M") as having been detected and sampled in the presence of motion. Alternatively, the presence of motion at the various lines / points can be detected by post-processing using a motion detection processor 540, as discussed in more detail below. In some embodiments, the motion detection processor 540 utilizes the navigator signal 545 to control or assist in the detection of motion.

[0042] The k-space pre-processor 510 (including the motion identification / rejection circuit 520) operates on the received set of k-space data 505 to generate a motion-removed set of k-space data 525 (including, but not limited to, from the first set of motion-corrupted k-space data 507). As shown, points with motion in the first set of motion-corrupted k-space data 507 are removed (and replaced with an "X" because they are indicated to have had motion during acquisition). PMSx (and associated points) are also removed (and replaced with an "X" because they are indicated to have had motion during acquisition). PSy (and associated points) are not indicated to have had motion during acquisition, but were determined to have been acquired in the presence of motion, e.g., using the navigator signal. Similarly, lines from the first set of motion-corrupted k-space data 507 can be removed due to motion after the k-space image data was acquired if the detected post-sampling was corrupted by motion. The resulting motion-removed set of k-space data 525 is undersampled.

[0043] The output of the motion identification / rejection circuit 520 is processed by a data extractor (DE) 526A circuit to generate a subset of the motion-removed set of k-space 525 (as a central set of motion-removed k-space data 527). The central set of motion-removed k-space data 527 is provided to circuitry for motion correction 530. The amount of data processed as the central set of motion-removed k-space data 527 may be a fixed size (e.g., all of the ACS data) or variable (e.g., 80%, 90%, or 95%, by calculating the data size required to represent a threshold amount of the total energy of the k-space data 505). In some embodiments, a second data extractor 526B extracts the first set of motion-corrupted k-space data 507 independently of the motion identification / rejection circuit 520.

[0044] The central set of motion-removed k-space data 527 is subjected to motion correction 530 to generate a first set of motion-corrected k-space data 535 (as in step 420 of FIG. 4 ) (potentially also using the first set 507 extracted from the motion-corrupted k-space data). The first set of motion-corrected k-space data 535 can then be used by a correction data generator 565 to generate correction data 570 for use as part of an undersampled reconstruction 580.

[0045] As shown in FIG. 5A, the first set of motion-corrected k-space data 535 replaces a portion of the motion-removed set of k-space data 525 to generate undersampled and motion-corrected combined k-space data 575. The undersampled and motion-corrected k-space data 575 then undergoes undersampled reconstruction 580 (using correction data 570) to generate a spatial image (as in step 420 of FIG. 4). The correction data may be generated, for example, by correction data generator 565A and may be in the form of a sensitivity map 570A (such as an ESPIRiT map) as shown in FIG. 5B. In such cases, the undersampled reconstruction may be implemented as a compressed sensing (CS) reconstruction 580A (FIG. 5B) or a deep learning (DLR) reconstruction 580C (FIG. 5D), with the image being generated based on the ESPIRiT map and the CS reconstruction 580A or the DLR reconstruction 580C, respectively. Alternatively, the correction data may be generated, for example, by correction data generator 565B and may be in the form of GRAPPA weight set 570B as shown in FIG. 5C. In such a case, the undersampled reconstruction may be implemented as GRAPPA reconstruction 580B. Undersampled reconstruction 580 may itself include motion compensation processing on undersampled and motion-compensated combined k-space data 575 (e.g., including first set of motion-compensated k-space data 535, which serves as ACS data).

[0046] As discussed above, the motion detection processor 540 can be used to facilitate motion identification and / or rejection (line-by-line or for all lines of an imaging shot). Navigator data can be acquired during every imaging shot, and the navigator can be a 3D volume, a 2D image, or a 1D signal. The navigator can be acquired from a variety of different sources, such as (1) non-imaging k-space echoes inserted into the pulse sequence, (2) respiratory bellows, (3) an ECG for cardiac motion, (4) a camera with or without external markers, and / or (5) pilot-tone-based motion detection. The navigator information can be used to detect motion in both the ACS and imaging data.

[0047] As shown in FIGS. 6A and 6B, one method for determining motion utilizes navigator correlation information (e.g., as represented in a plot) resulting from non-imaging k-space echoes (e.g., directed at readout (RO) and phase encoding (PE)) collected on a shot-by-shot basis. In one such embodiment, an inverse Fourier transform of the navigator k-space is obtained. Then, a correlation (correlation plot) between each navigator signal and the reference navigator is obtained, as shown in FIG. 6A. Shots with correlation values ​​below a threshold (α) are motion-intensive and therefore rejected, as shown in FIG. 6B. The threshold can be determined in various ways, including empirically. For example, an absolute threshold can be used to treat shots with correlations below α (e.g., 0.999) as motion-corrupted (such as shots 1-6 in FIG. 6B). Alternatively, a standard deviation can be used to treat shots with correlations below (μ-α×σ) as motion-corrupted. Here, the mean (μ) and standard deviation (σ) can be calculated using correlations of all shots or a subset of shots. To avoid excessive removal of corrupted portions before reconstructing the central portion (e.g., the ACS portion), a limit may be placed on the total number of shots that can be designated as corrupted. For example, the total number of shots that can be treated as corrupted may be limited to be less than α × the total number of shots. In such an embodiment, if there are more shots that can be treated as motion corrupted than (α × the total number of shots), the shots to be treated as motion corrupted may be selected randomly by marking them in the order in which they are discovered, or by selecting the (α × the total number of shots) shots with the lowest correlation values. Furthermore, to avoid large gaps in k-space after shot rejection before reconstructing the central portion (e.g., the ACS portion), a limit may be placed on the maximum gap size after undersampling that can occur in k-space due to shot rejection. These thresholding methods may be used in combination.

[0048] FIG. 7A shows a second exemplary correlation plot illustrating motion in the ACS of k-space imaging data from a T2w FSE brain scan. Based on the correlation values, shots 7, 8, and 9 (and their corresponding lines) are rejected due to the presence of motion. FIG. 7B shows the ACS (sum of squares) image corresponding to the correlation information for the scan of FIG. 7A, which contains motion corruption in the ACS information, indicated by the arrows in FIG. 7B. Motion is also evident, as indicated by the arrows in FIG. 7D, which represents the ACS data from the ACS k-space coil. After ACS motion correction processing (step 420) using the GRAPPA method, described in more detail below, the ACS image and ACS k-space coil data are improved, as shown in FIGS. 7C and 7E, respectively.

[0049] Motion detection can also be performed using deep learning-based (DL-based) classification. Deep learning can detect motion-corrupted captured shots by training a DL network with simulated motion using correlation plots or raw navigator signals. In one such embodiment, a DL network is trained by (1) selecting navigator data that is corrupted by rigid body motion, (2) selecting shots that will be corrupted by through-plane motion (e.g., because the simulated intensity changes with the navigator signal), and (3a) if DL is applied directly to the navigator signal, the navigator signal is input to the DL network and a list of navigator signals with motion is output from the DL network, or (3b) if DL is applied to an association plot, the association plot is input to the DL network and a list of motion-corrupted shots is output from the DL network. In step (1), translation and rotation can be simulated for 3D and 2D navigators, and translation in the readout direction can be simulated for a 1D navigator.

[0050] As discussed above with respect to step 420 and motion correction 530 of FIGS. 5A and 5B, motion correction can be applied to the first set 507 of motion-corrupted k-space data. FIGS. 8A-8C are internal block diagrams of motion correction 530 according to one embodiment at three different iterations of GRAPPA reconstruction. FIG. 8A corresponds to the initial setup (iteration 0) of the GRAPPA reconstruction, where the first set 507 of motion-corrupted k-space data is used for the initial estimation of GRAPPA weights 8300 for the undersampled ACS reconstruction (ACS(0)). Furthermore, the central set 527 of motion-removed (undersampled) k-space data is treated as the GRAPPA input, denoted as Z in the following equation: Then, in subsequent iterations (where n increments from 1 to N), the current GRAPPA weights 830 are calculated according to the weighted combination as in equation (1) below: n The new weight is 830 n+1 is replaced by

[0051]

number

[0052] As shown in Figure 8C, after the last iteration (n = N), ACS(N) is equal to the first set of motion-corrected k-space data 535. The iterative GRAPPA process as described above is described in Zhao, T., & Hu, X. (2008), Iterative GRAPPA (iGRAPPA) for improved parallel imaging reconstruction Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine, 59(4), 903-907, the contents of which are incorporated herein by reference.

[0053] As shown in FIG. 9 , in an alternative embodiment, motion correction 530 is performed in a multi-step process. In an initial step, as in the initial iteration of FIG. 8A , a first set of motion-corrupted k-space data 507 and a central set of motion-removed (undersampled) k-space data 527 are applied to a GRAPPA reconstruction to generate an initial GRAPPA reconstruction of the first set of k-space data. The initial GRAPPA reconstruction of the first set of k-space data is then used with the first set of motion-corrupted k-space data 507 to generate an initial RAKI reconstruction of the first set of k-space data. A total of N iterative reconstructions are then performed to generate a first set of motion-corrected k-space data 535. In each iteration of the iterative RAKI process, the weights of a convolutional neural network (CNN) are refined using the current iteration of the first set of k-space data (previously corrected). The iterative RAKI process is described in Dawood, P. et al., Iterative training of robust k-space interpolation networks for improved image reconstruction with limited scan-specific training samples. Magnetic Resonance in Medicine, 89(2), 812-827 (2023), the contents of which are incorporated herein by reference.

[0054] In an alternative embodiment shown in Figure 10, the system 500 does not generate undersampled and motion-corrected combined k-space data 575 as in Figure 5A. Instead, the motion-removed set of k-space 525 is directly applied to an undersampled reconstruction 580 (which may include motion compensation) using the corrected data 570.

[0055] As shown in Figures 11A-11C, motion correction of the first set 507 of motion-corrupted k-space data before undersampled reconstruction (or full reconstruction) can improve image quality using CS reconstruction. Figure 11A shows an image generated from motion-corrupted k-space data from a T2W FSE brain scan that underwent CS reconstruction. Figure 11B shows the resulting image when the first set 507 of motion-corrupted ACS k-space data was used in an undersampled reconstruction using compressed sensing (CS) reconstruction after shot rejection. Finally, as shown in Figure 11C, the best image quality was achieved by performing a separate motion correction on the ACS k-space data corresponding to Figure 11A before performing undersampled CS reconstruction on the motion-corrected combined k-space data and the motion-corrected combined ACS k-space data and non-ACS k-space data.

[0056] As shown in Figures 12A-12C, motion correction of the first set of motion-corrupted k-space data 507 prior to undersampled reconstruction (or fully sampled reconstruction) can improve image quality, even when GRAPPA reconstruction is also used. Figure 12A shows an image (generated from the same motion-corrupted k-space data from the T2W FSE brain scan of Figure 11A) using GRAPPA reconstruction. Figure 12B shows the resulting image when the first set of motion-corrupted ACS k-space data 507 is used for undersampled reconstruction using GRAPPA reconstruction after shot rejection. Finally, as shown in Figure 12C, the best image quality was achieved by performing motion correction on the ACS k-space data corresponding to Figure 12A before performing undersampled GRAPPA reconstruction on the motion-corrected combined k-space data and motion correction on the motion-corrected combined ACS k-space data and non-ACS k-space data.

[0057] The methods and systems described herein can be implemented using numerous technologies, but generally relate to imaging devices and processing circuitry for performing the processes described herein. In one embodiment, the processing circuitry (e.g., image processing circuitry and controller circuitry) is implemented as one or a combination of an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a generic array of logic (GAL), a programmable array of logic (PAL), one-time programmable logic gates (e.g., using fuses), or reprogrammable logic gates. The processing circuitry can also include a computer processor having internal and / or external non-volatile computer-readable memory (e.g., RAM, SRAM, FRAM, PROM, EPROM, and / or EEPROM) that stores computer instructions (binary executable instructions and / or interpreted computer instructions) that control the computer processor to perform the processes described herein. The computer processor circuitry may implement a single processor or multiple processors, each supporting a single thread or multiple threads and each having a single core or multiple cores.

[0058] Embodiments of the present disclosure may be as described in the supplement below.

[0059] (1) A method of image processing, including, but not limited to, (a) receiving k-space data acquired by scanning an object with a magnetic resonance imaging device, the k-space data including a first set of motion-corrupted k-space data and a second set of k-space data different from the first set of motion-corrupted k-space data; (b) generating motion-corrected data based on the first set of motion-corrupted k-space data and information indicating whether the object moved while being scanned; and (c) generating an image based on the second set of k-space data and the motion-corrected data.

[0060] (2) The method described in (1), wherein the k-space data corresponding to the motion of the object in the first set of motion-corrupted k-space data is not used to generate the motion-corrected data.

[0061] (3) A method according to (2), wherein the motion-corrected data is generated based on motion-corrected k-space data generated by applying an iterative GRAPPA process to k-space data in the first set of motion-corrupted k-space data that does not correspond to motion of the object.

[0062] (4) A method according to (2), wherein the motion-corrected data is generated based on motion-corrected k-space data generated by applying an iterative GRAPPA process and an iterative RAKI process to k-space data in the first set of motion-corrupted k-space data that does not correspond to the motion of the object.

[0063] (5) A method according to any one of (1) to (4), wherein the first set of motion-corrupted k-space data is at least one of undersampled k-space data and data acquired by parallel imaging.

[0064] (6) The method according to any one of (1) to (5), wherein the first set of motion-corrupted k-space data is auto-calibration signal (ACS) data.

[0065] (7) A method according to any one of (1) to (6), wherein generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the motion-corrected data and data in the second set of k-space data that is not motion-corrupted.

[0066] (8) The method according to (7), wherein the motion-corrected data is sensitivity information indicating the sensitivity of each of a plurality of coils for receiving magnetic resonance signals from the object, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the sensitivity information and data in the second set of k-space data that is not motion-corrupted.

[0067] (9) The method of (7), wherein the motion-corrected data is sensitivity information for each of a plurality of coils that receive magnetic resonance signals from the object, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the sensitivity information and an image generated from data in the second set of k-space data that is not motion-corrupted.

[0068] (10) The method of (7), wherein the motion-corrected data is a set of GRAPPA weights, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the set of GRAPPA weights and data in the second set of k-space data that is not motion-corrupted.

[0069] (11) The method of (7), wherein the motion-corrected data is a set of GRAPPA weights, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the set of GRAPPA weights and an image generated from data in the second set of k-space data that is not motion-corrupted.

[0070] (12) The method according to (1) or (2), wherein the second set of k-space data is undersampled k-space data.

[0071] (13) The method of (12), wherein the motion-corrected data is an ESPIriT map, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the ESPIriT map and the second set of undersampled k-space data.

[0072] (14) The method of (12), wherein the motion-corrected data is an ESPIriT map, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the ESPIriT map and an image generated from the undersampled second set of k-space data.

[0073] (15) The method according to (12), wherein the motion-corrected data is a set of GRAPPA weights, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on k-space data interpolated by the set of GRAPPA weights and the second set of undersampled k-space data.

[0074] (16) The method according to (12), wherein the motion-corrected data is a set of GRAPPA weights, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on k-space data interpolated by the set of GRAPPA weights and an image generated from the second set of undersampled k-space data.

[0075] (17) The method according to (12), wherein the motion-corrected data is sensitivity information indicating the sensitivity of each of a plurality of coils that receive magnetic resonance signals from the object, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the sensitivity information and the undersampled second set of k-space data.

[0076] (18) The method according to (12), wherein the motion-corrected data is sensitivity information indicating the sensitivity of each of a plurality of coils that receive magnetic resonance signals from the object, and generating the image based on the second set of k-space data and the motion-corrected data includes, but is not limited to, generating the image based on the sensitivity information and an image generated from the undersampled second set of k-space data.

[0077] (19) The method according to any one of (1) to (18), wherein the information indicating whether the object has moved while scanning the object is detected using a navigator signal.

[0078] (20) An apparatus for performing image processing, comprising a processing circuit that performs any one of steps (1) to (19).

[0079] (21) A non-transitory computer-readable storage medium storing computer-readable instructions that, when executed by a computer, cause the computer to perform any one of the image processing methods (1) to (19).

[0080] Thus, the foregoing discussion discloses and describes merely exemplary embodiments of the present disclosure. As will be understood by those skilled in the art, the present disclosure may be embodied in other specific forms without departing from the spirit thereof. Accordingly, the disclosure of the present disclosure is intended to illustrate, but not limit, the scope of the disclosure as well as the other claims. Such disclosure, including any readily discernible variations of the teachings described herein, will in part define the scope of the foregoing claim terms so as not to divulge inventive subject matter. [Explanation of symbols]

[0081] 1 MRI machine 10 Static magnetic field magnet 11 Gradient magnetic field coil 12 WB coil 20 RF coils 31, 31x, 31y, 31z Gradient coil power supply 32 RF receiver 33 RF transmitter 34 Sequence Controller 36 Coil selection circuit 40 Console 41 memory 42 Display 43 Input Interface 45 Processing circuit 50 berths 51 units 52 Bed body 100 Mounting Stand 300 Control Cabinet 400 ways 500 Systems 505 k-space data sets 507,535 1st set 510 k-space preprocessor 520 Motion Identification / Rejection Circuit 525 sets 526B Second Data Extractor 527 Central Set 530 Motion Compensation 540 Motion Detection Processor 545 Navigator Signal 565, 565A, 565B Correction Data Generator 570 Correction Data 570A Sensitivity Map 570B GRAPPA Weight Set 575 combined k-space data 580 Reconfiguration 580A CS reconfiguration 580B GRAPPA Reconfiguration 580C DLR Reconfiguration

Claims

1. receiving k-space data acquired by scanning an object with a magnetic resonance imaging device, the k-space data including a first set of motion-corrupted k-space data and a second set of k-space data different from the first set of motion-corrupted k-space data; generating motion-corrected data based on the first set of motion-corrupted k-space data and information indicating whether the object moved while scanning the object; generating an image based on the second set of k-space data and the motion corrected data; An image processing method comprising:

2. the k-space data in the first set of motion-corrupted k-space data corresponding to motion of the object is not used to generate the motion-corrected data. The image processing method according to claim 1 .

3. the motion-corrected data is generated based on motion-corrected k-space data generated by applying an iterative GRAPPA process to k-space data of the first set of motion-corrupted k-space data that does not correspond to motion of the object. The image processing method according to claim 2 .

4. the motion-corrected data is generated based on motion-corrected k-space data generated by applying at least one of an iterative GRAPPA process or an iterative RAKI process to k-space data of the first set of motion-corrupted k-space data that does not correspond to motion of the object; The image processing method according to claim 2 .

5. the first set of motion-corrupted k-space data is at least one of undersampled k-space data and data acquired by parallel imaging; The image processing method according to claim 1 .

6. the first set of motion-corrupted k-space data is auto-calibration signal (ACS) data; The image processing method according to claim 1 .

7. generating the image based on the second set of k-space data and the motion-corrected data includes generating the image based on the motion-corrected data and data in the second set of k-space data that is not motion corrupted. The image processing method according to claim 1 .

8. the motion correction data is sensitivity information indicating sensitivity of each of a plurality of coils receiving magnetic resonance signals from the object; generating the image based on the second set of k-space data and the motion-corrected data includes generating the image based on the sensitivity information and data in the second set of k-space data that is not motion-corrupted. The image processing method according to claim 7.

9. the motion-compensated data is a set of GRAPPA weights; generating the image based on the second set of k-space data and the motion corrected data includes generating the image based on the set of GRAPPA weights and data in the second set of k-space data that is not motion corrupted. The image processing method according to claim 7.

10. the second set of k-space data is undersampled k-space data. The image processing method according to claim 1 .

11. the motion-corrected data is an ESPIRiT map; generating the image based on the second set of k-space data and the motion corrected data includes generating the image based on the ESPIRiT map and the second set of undersampled k-space data. The image processing method according to claim 10.

12. the motion-compensated data is a set of GRAPPA weights; generating the image based on the second set of k-space data and the motion corrected data includes generating the image based on k-space data interpolated with the set of GRAPPA weights and the second set of undersampled k-space data. The image processing method according to claim 10.

13. the motion correction data is sensitivity information indicating sensitivity of each of a plurality of coils receiving magnetic resonance signals from the object; generating the image based on the second set of k-space data and the motion corrected data includes generating the image based on the sensitivity information and the undersampled second set of k-space data. The image processing method according to claim 10.

14. the information indicating whether the object moved while scanning the object is detected using a navigator signal; The image processing method according to claim 1 .

15. receiving k-space data acquired by scanning an object with a magnetic resonance imaging device, the k-space data including a first set of motion-corrupted k-space data and a second set of k-space data different from the first set of motion-corrupted k-space data; generating motion-corrected data based on the first set of motion-corrupted k-space data and information indicating whether the object moved while scanning the object; generating an image based on the second set of k-space data and the motion corrected data; 1. An image processing device comprising: a processing circuit configured to perform

16. The image processing apparatus of claim 15 , wherein the k-space data in the first set of motion-corrupted k-space data that corresponds to motion of the object is not used to generate the motion-corrected data.

17. the motion-corrected data is generated based on motion-corrected k-space data generated by applying an iterative GRAPPA process to k-space data of the first set of motion-corrupted k-space data that does not correspond to motion of the object. The image processing device according to claim 16.

18. the motion-corrected data is generated based on motion-corrected k-space data generated by applying at least one of an iterative GRAPPA process or an iterative RAKI process to k-space data of the first set of motion-corrupted k-space data that does not correspond to motion of the object; The image processing device according to claim 16.

19. the first set of motion-corrupted k-space data is at least one of undersampled k-space data and data acquired by parallel imaging; The image processing device according to claim 15.

20. the first set of motion-corrupted k-space data is auto-calibration signal (ACS) data; The image processing device according to claim 15.

Images