System and method for motion correction of propeller magnetic resonance images

Abstract

The present invention provides a magnetic resonance (MR) imaging method (550) for correcting motion in a pre-corrected MR image of a subject. The method includes applying a pulse sequence via an MR system having a k-space trajectory of a blade rotating in k-space. The method also includes acquiring k-space data of a three-dimensional (3D) imaging volume of the subject, the k-space data of which corresponds to the pre-corrected MR image and is acquired via the pulse sequence. The method further includes receiving (554) 3D MR calibration data of a 3D calibration volume, wherein the 3D calibration volume is greater than or equal to the 3D imaging volume; jointly estimating (556) rotation and translation in the pre-corrected MR image based on the k-space data of the 3D imaging volume and the calibration data; correcting (558) motion in the pre-corrected image based on the estimated rotation and the estimated translation; and outputting (560) the motion-corrected image.

Description

Background Technology

[0001] The field of this disclosure relates generally to systems and methods for motion correction, and more specifically, to systems and methods for motion correction of magnetic resonance (MR) images acquired via PROPELLER (periodic rotating overlapping parallel lines with enhanced reconstruction).

[0002] Magnetic resonance imaging (MRI) has proven useful for the diagnosis of many diseases. MRI provides detailed images of soft tissues, abnormal tissues (such as tumors), and other structures that cannot be easily imaged by other imaging modalities such as computed tomography (CT). Furthermore, MRI operates without exposing the patient to the ionizing radiation experienced in modalities such as CT and X-rays.

[0003] Although the PROPELLER sampling scheme is relatively insensitive to motion compared to Cartesian sampling, it still requires motion correction to reduce motion-introduced artifacts. Known motion correction methods are disadvantageous in some respects and require improvement. Summary of the Invention

[0004] In one aspect, a magnetic resonance (MR) imaging method for correcting motion in a pre-corrected MR image of a subject is provided. The method includes applying a pulse sequence via an MR system, the pulse sequence having a k-space trajectory of a blade rotating in k-space, the blade comprising multiple views. The method also includes acquiring k-space data of a three-dimensional (3D) imaging volume of the subject, the k-space data of which corresponds to the pre-corrected MR image and is acquired via the pulse sequence. The method further includes receiving 3D MR calibration data of a 3D calibration volume, wherein the 3D calibration volume is greater than or equal to the 3D imaging volume; jointly estimating rotation and translation in the pre-corrected MR image based on the k-space data of the 3D imaging volume and the calibration data; correcting motion in the pre-corrected image based on the estimated rotation and the estimated translation; and outputting a motion-corrected image.

[0005] On the other hand, a motion correction system is provided for correcting motion in a pre-corrected MR image of a subject. The system includes a motion correction computing device comprising at least one processor communicating with at least one memory device. The at least one processor is programmed to receive k-space data of a 3D imaging volume of the subject, the k-space data corresponding to the pre-corrected MR image and acquired via a pulse sequence having a k-space trajectory of a blade rotating in k-space, the blade including multiple views. The at least one processor is also programmed to receive 3D MR calibration data of a 3D calibration volume, wherein the 3D calibration volume is greater than or equal to the 3D imaging volume. The at least one processor is further programmed to jointly estimate rotation and translation in the pre-corrected MR image based on the k-space data of the 3D imaging volume and the calibration data, correct motion in the pre-corrected image based on the estimated rotation and the estimated translation, and output a motion-corrected image. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of an exemplary magnetic resonance imaging (MRI) system.

[0007] Figure 2A This is a schematic diagram of the k-space trajectory of the PROPELLER (periodic rotating overlapping parallel lines with enhanced reconstruction) sampling scheme.

[0008] Figure 2B yes Figure 2A A schematic diagram of the blade in the sampling scheme shown.

[0009] Figure 3A This is a comparison of images acquired using PROPELLER and images acquired using Descartes fast spin echo.

[0010] Figure 3B This is a comparison between images acquired using PROPELLER and images acquired using echo-plane imaging.

[0011] Figure 4A This is a flowchart of a known motion correction method for images acquired using PROPELLER.

[0012] Figure 4B Images without motion correction and those with Figure 4A A comparison of known motion-corrected images is shown.

[0013] Figure 4C It shows having Figure 4A The image shown is of adjacent slices with known motion correction.

[0014] Figure 4D Images with edge slices without motion correction and Figure 4A The comparison shown is of images with edge slices having known motion correction.

[0015] Figure 5A This is a schematic diagram of an exemplary motion correction system.

[0016] Figure 5B This is a flowchart of an exemplary method for motion correction.

[0017] Figure 6 This is a flowchart of an exemplary method for jointly estimating rotation and translation.

[0018] Figure 7 Through Figure 4A The image shown is subjected to motion correction by the known method and through Figures 5A to 6 A comparison of motion-corrected images using the systems and methods shown.

[0019] Figure 8 Through Figure 4A The image shown is subjected to motion correction by the known method and through Figures 5A to 6 Another comparison of motion-corrected images using the system and method shown.

[0020] Figure 9 This is a block diagram of an exemplary computing device. Detailed Implementation

[0021] This disclosure includes systems and methods for motion correction of magnetic resonance (MR) images acquired from a subject using the PROPELLER (periodic rotating overlapping parallel lines with enhanced reconstruction) technique. As used herein, motion correction, or correcting motion of an image, is the process of transforming an image back to a state where no motion occurs, or reducing or removing motion-induced artifacts. The subject is a human, animal, or phantom. Motion correction of image slices acquired via a 2D PROPELLER is described herein only as an example. The systems and methods disclosed herein can be applied to motion correction of images acquired in 3D via a PROPELLER (such as images acquired via a PROPELLER in multiple slice encoding steps). Methodological aspects will be apparent in part and discussed explicitly in part in the following description.

[0022] In magnetic resonance imaging (MRI), the subject is placed in a magnet. When the subject is in a magnetic field generated by the magnet, the magnetic moments of nuclei, such as protons, attempt to align with the magnetic field, but precess around the field in a random order at the Larmor frequency of the nucleus. The magnetic field of the magnet is called B0 and extends longitudinally, or in the z-direction. During the acquisition of MRI images, a magnetic field in the xy-plane and close to the Larmor frequency (called the excitation field B1) is generated by a radio frequency (RF) coil and can be used to rotate or “tilt” the net magnetic moment Mz of the nucleus from the z-direction toward the transverse or xy-plane. After the excitation signal B1 terminates, the nucleus emits a signal, which is called the MR signal. To generate an image of the subject using the MR signal, magnetic field gradient pulses (Gx, Gy, and Gz) are used. The gradient pulses are used to scan the reverse direction of space or distance through k-space, spatial frequency, and space. There is a Fourier relationship between the acquired MR signal and the image of the subject, so the image of the subject can be derived by reconstructing the MR signal.

[0023] Figure 1 A schematic diagram of an exemplary MRI system 10 is shown. In an exemplary embodiment, the MRI system 10 includes a workstation 12 having a display 14 and a keyboard 16. The workstation 12 includes a processor 18, such as a commercially available programmable machine running a commercially available operating system. The workstation 12 provides an operator interface that allows scanning protocols to be input into the MRI system 10. The workstation 12 is coupled to a pulse sequence server 20, a data acquisition server 22, a data processing server 24, and a data storage server 26. The workstation 12 and each of the servers 20, 22, 24, and 26 communicate with each other.

[0024] In an exemplary embodiment, the pulse sequence server 20 responds to instructions downloaded from workstation 12 to operate the gradient system 28 and the radio frequency (“RF”) system 30. The instructions are used to generate gradient waveforms and RF waveforms in the MR pulse sequence. The RF coil 38 and the gradient coil assembly 32 are used to execute the prescribed MR pulse sequence. The RF coil 38 is shown as a whole-body RF coil. The RF coil 38 may also be a local coil that can be placed near the anatomical structure to be imaged, or a coil array comprising multiple coils.

[0025] In an exemplary embodiment, a gradient waveform for performing a defined scan is generated and applied to a gradient system 28, which excites the gradient coils in the gradient coil assembly 32 to generate a magnetic field gradient G for frequency encoding, phase encoding, and slice selection / encoding of the MR signal. x G y and G z The gradient coil assembly 32 forms part of the magnet assembly 34, which also includes a polarized magnet 36 and an RF coil 38.

[0026] In an exemplary embodiment, RF system 30 includes an RF transmitter for generating RF pulses used in an MR pulse sequence. The RF transmitter responds to a scan scheme and orientation from pulse sequence server 20 to generate RF pulses with desired frequency, phase, and pulse amplitude waveforms. The generated RF pulses can be applied by RF system 30 to RF coil 38. The responsive MR signal detected by RF coil 38 is received by RF system 30 and amplified, demodulated, filtered, and digitized under the direction of commands generated by pulse sequence server 20. RF coil 38 is described as both a transmitter and a receiver coil, such that RF coil 38 transmits RF pulses and detects MR signals. In one embodiment, MRI system 10 may include a transmitter RF coil for transmitting RF pulses and a separate receiver coil for detecting MR signals. A transmission channel of RF system 30 can be connected to the RF transmission coil, and a receiver channel can be connected to a separate RF receiver coil. Typically, the transmission channel is connected to the whole-body RF coil 38, and each receiver segment is connected to a separate local RF coil.

[0027] In an exemplary embodiment, the RF system 30 further includes one or more RF receiver channels. Each RF receiver channel includes an RF amplifier that amplifies the MR signal received by the RF coil 38 to which the channel is connected; and a detector that detects and digitizes the I quadrature components and Q quadrature components of the received MR signal. The magnitude of the received MR signal can then be determined as the square root of the sum of the squares of the I and Q components, as shown in equation (1) below:

[0028]

[0029] Furthermore, the phase of the received MR signal can also be determined as shown in equation (2) below:

[0030]

[0031] In an exemplary embodiment, digitized MR signal samples generated by RF system 30 are received by data acquisition server 22. Data acquisition server 22 can operate in response to instructions downloaded from workstation 12 to receive real-time MR data and provide buffer storage so that no data is lost due to data overflow. In some scans, data acquisition server 22 simply transmits the acquired MR data to data processing server 24. However, in scans where information derived from the acquired MR data is needed to control further execution of the scan, data acquisition server 22 is programmed to generate the required information and transmit it to pulse sequence server 20. For example, during a pre-scan, MR data is acquired and used to calibrate a pulse sequence performed by pulse sequence server 20. Additionally, navigator signals can be acquired during the scan and used to adjust operating parameters of RF system 30 or gradient system 28, or to control the view sequence for sampling k-space.

[0032] In an exemplary embodiment, the data processing server 24 receives MR data from the data acquisition server 22 and processes the MR data according to instructions downloaded from the workstation 12. Such processing may include, for example, performing a Fourier transform on the raw k-space MR data to produce a two-dimensional or three-dimensional image, applying filters to the reconstructed image, generating a functional MR image, and calculating a motion or flow image.

[0033] In an exemplary embodiment, the image reconstructed by the data processing server 24 is transmitted back to workstation 12 and stored there. In some embodiments, the real-time image is stored in a database storage cache. Figure 1 (Not shown in the image) Real-time images can be output from the database storage cache to the operator's display 14 or a display 46 located near the magnet assembly 34 for use by the attending physician. Batch-processed images or selected real-time images can be stored on disk storage 48 or a host database in the cloud. When such images have been reconstructed and transferred to the storage device, the data processing server 24 notifies the data storage server 26. The operator can use workstation 12 to archive images, generate films, or send images to other facilities via a network.

[0034] The systems and methods disclosed herein are used to reduce or remove motion artifacts in images acquired via PROPELLER. MRI is unique compared to other imaging modalities because MRI signals are represented by complex numbers, rather than scalar or real numbers. Therefore, the image value of each image pixel includes both magnitude and phase. Composite MR images can be reconstructed based on I-orthogonal MR signals and Q-orthogonal MR signals using procedures such as Fourier transform.

[0035] Figures 2A to 2B The PROPELLER k-space sampling scheme is shown. Figure 2A This is a schematic diagram of the k-space trajectory 200 of the PROPELLER sampling scheme. Figure 2B The blade 202 in the PROPELLER sampling scheme is shown. The PROPELLER sampling scheme is a scheme for sampling k-space using a PROPELLER pulse sequence, in which k-space is sampled by rotating blade 202, where each blade 202 has multiple views 206. Images acquired by PROPELLER can be referred to as PROPELLER images. In MR, the pulse sequence is a sequence of RF pulses, gradient pulses, and data acquisition applied by the MRI system 10 when acquiring MR signals. As described above, in MRI, MR signals are acquired by sampling k-space using gradients Gx, Gy, and Gz, which correspond to kx / frequency encoding direction, ky / phase encoding direction, and kz / slice encoding direction, respectively. MR images can be reconstructed from MR signals using Fourier transform. Typically, k-space is sampled along a Cartesian grid, where the k-space sampling points form a linear pattern. The advantage of Cartesian sampling is that the data points are regularly spaced and can be directly placed in standard array processors designed for Fast Fourier Transform (FFT) calculations. The drawback of Cartesian sampling is its sensitivity to motion.

[0036] The PROPELLER pulse sequence does not sample the k-space along a Cartesian grid. Instead, it samples the k-space in a radial pattern, where the k-space lines radially cross the central region of the k-space and is classified as a radial sampling scheme. Compared to standard radial acquisition (which samples a line after the RF excitation pulse), in the PROPELLER sampling scheme, the blade 202, comprising multiple views or k-space lines 206, is sampled after the RF excitation pulse via fast spin echo sampling or other methods (such as gradient echo). The number of views 206 in a blade 202 is typically between 16 and 32. The views 206 in the blade 202 are generally parallel to each other, with the central view 206-c of the blade 202 potentially crossing the center 208 of the k-space. After sampling the blade 202 at a certain angle, the blade 202 is rotated, at which point a second set of data, a third set, and so on, is acquired. This process continues until MR data for the entire k-space circle has been collected.

[0037] The main benefit of radial sampling is its relative insensitivity to motion. Unlike Cartesian sampling, radial sampling does not have a fixed frequency and phase encoding direction. Noise from moving anatomical structures does not propagate as discrete ghosting, blurring, or artifacts along a single phase encoding direction. In radial sampling, this noise is dispersed throughout the image. Furthermore, because the central region of k-space is oversampled, all radial lines contribute equally to the image, including relatively strong signals from the central region of k-space, unlike Cartesian sampling which samples the central region of k-space only on a few lines. Therefore, motion during one or a few radial lines is unlikely to significantly degrade image quality. Additionally, in PROPELLER, the central region R is sampled through all blades 202. This redundancy can be used to estimate motion and thus for motion correction.

[0038] Figure 3A and Figure 3B Images 302, 304, and 305 acquired using PROPELLER are shown compared to images 306, 308, and 309 acquired using Cartesian sampling over the same anatomical region with the same matrix size and field of view (FOV). Image 306 was acquired using Cartesian fast spin echo (FSE). Images 308 and 309 were acquired using echo-plane imaging (EPI). Images 305 and 309 are diffusion-weighted images. PROPELLER image 302 is less blurred and more uniform than FSE image 306. Similarly, PROPELLER images 304 and 305 have less image distortion and are more uniform than EPI images 308 and 309.

[0039] Figures 4A to 4D The image shows a known motion correction method 400, as well as images without motion correction and with motion correction performed by method 400. The known method 400 assumes that motion parameters can be estimated from k-space data. Method 400 includes estimating and correcting the rotation of 402MR data using mass k-space data. Mass k-space data M from the central region R of each blade 202 (Fig. 2) is shown. n The data are gridded onto R and averaged together to form a reference k-space data M. A Each M n Rotated according to a series of angles and meshed onto R for each angle. For each blade 202, M n With M A The correlation was calculated as a function of the rotation angle. The calculated correlation in M n Rotate to M AThe correlation (as a function of rotation angle) is maximized at this point. It is fitted to a second-order polynomial function, and the peak of the polynomial function is the estimated rotation angle of blade 202. Once the rotation angles of all blades 202 are estimated, the coordinates of each blade are rotated according to that blade's estimated rotation angle. In other words, the rotational portion of the motion correction is completed.

[0040] Then, the composite k-space data D with rotationally corrected coordinates at the central region R is used. n To estimate and correct the 404 translation. Similar to rotation estimation, a composite rotation correction k-space data D is performed at the central region R of all blades 202. n Take the average value to derive the average composite data D. A When estimating the translation, D... A Composite conjugated D A * Multiply by D n The product is then subjected to a Fourier transform. The peak value of the Fourier transform product is determined, and a three-point parabolic fit is performed on the peak value in the x and y directions. The position of the vertex of this parabola is the estimated translation in the x and y directions, and the corresponding linear phase is removed from the data collected by blade 202 to correct the translational portion of the motion, since the translation in the actual or image space xy corresponds to a linear phase change in k-space kx-ky. At this point, the k-space data is subjected to motion correction, or rotation and translation correction. This motion correction is performed slice by slice. That is, motion correction only corrects in-plane motion, i.e., motion occurring within a slice.

[0041] After performing motion correction 400, correlation thresholding can be performed to discard or apply low weights to k-space data from blade 202 corresponding to significant through-plane motion, uncorrected non-rigid volume motion, or other factors that will produce artifacts in the final reconstructed image. The final image is reconstructed using the correlation-weighted k-space data from blade 202 by meshing the correlation-weighted k-space data and applying a Fourier transform.

[0042] Method 400 assumes that the motion is primarily rigid within a plane and that the motion can be estimated based on phase and signal variations in k-space data. However, this assumption is not always correct. Therefore, Method 400 has several drawbacks. First, it does not correct for motion across the plane. Figure 4BThe image shows a coronal image 412 acquired by PROPELLER without using motion correction method 400 and a coronal image 414 based on the same k-space data but using motion correction 400. Although the motion-corrected image 414 has reduced motion artifacts 416 compared to the uncorrected image 412, motion artifacts 416 are still present in image 414 because method 400 only corrects in-plane motion artifacts. Furthermore, each slice is individually aligned, therefore slices can be misaligned. Figure 4C Axial images 418 and 420 of two adjacent slices (i.e., slices adjacent to each other) in the brain are shown, and a sagittal image 422 is generated based on a series of axial images including images 418 and 420. For Figure 4C The image shown is generated by acquiring a series of axial slices using a PROPELLER pulse sequence. Motion correction is performed on the axial images using method 400, and sagittal images are generated using the axial images to display images from different directions. In MR, to minimize crosstalk between the slice contours of adjacent slices in the slice-selective RF pulse, even-numbered and odd-numbered slices can be acquired in different traversals. The subject can move between acquisitions in different traversals. In motion correction 400, the individual axial images 418, 420 are aligned independently of each other. Therefore, the slices are misaligned, which appears as a jagged pattern 424 along the slice direction 423 in the sagittal image 422.

[0043] Third, because in Method 400, the rotation of 402 is estimated and corrected before the translation of 404 is estimated and corrected, Method 400 relies heavily on the robustness of rotation estimation and correction. For limbic slices of the brain, rotation estimation fails when the anatomical structure is relatively round and symmetrical. Figure 4D ). Figure 4D An axial image 426 based on k-space data acquired by PROPELLER without motion correction 400 is shown, along with an axial image 428 based on the same acquired k-space data but with motion correction 400. The image quality of the motion-corrected image 428 is significantly lower than that of the uncorrected image 426. For example, the motion-corrected image 428 has a greatly increased number of artifacts 416. More importantly, the rotation estimation is very inaccurate, rendering the estimated rotation unusable. That is, the rotation estimation and correction 402 in method 400 fail for edge slices.

[0044] Fourth, Method 400 is only effective for brain images and tends to fail when applied to PROPELLER images of other anatomical structures in the body because Method 400 does not correct for through-plane motion, and other anatomical structures in the body have more through-plane motion from respiratory and / or cardiac movements. Finally, for deviations from magnet 36 (see...) Figure 1Motion correction 400 often fails for isocentric (or off-center) imaging volumes or regions of the magnet 36. The magnetic field of system 10 at locations far from the isocentric of magnet 36 is not as uniform as at locations near or at the isocentric, and this non-uniformity manifests as phase variations in the k-space data. Because motion estimation and correction in method 400 are based on composite k-space data including phase information, method 400 often fails for off-center anatomical structures, such as the wrist, knee, or shoulder in musculoskeletal (MSK) imaging.

[0045] The systems and methods described herein address the problems associated with method 400. Instead of being based on k-space data, the motion correction systems and methods described herein derive motion parameters from images. The systems and methods described herein correct both in-plane and trans-plane motion in image space, rather than only correcting in-plane motion in k-space and discarding k-space data containing trans-plane motion and other undesirable factors. Trans-plane motion can be corrected via slice-to-volume registration. The systems and methods described herein are robust and extend the application of motion correction for PROPELLER images to edge slices of the brain, other anatomical structures in the body outside the brain, off-center imaging volumes, and diffusion-weighted imaging.

[0046] Figure 5A This is a schematic diagram of an exemplary motion correction system 500. In an exemplary embodiment, system 500 includes a motion correction computing device 502 configured to correct motion in MR images. The motion correction computing device 502 may be included in a workstation 12 of the MRI system 10, or may be included in a separate computing device that communicates with the workstation 12 via wired or wireless communication. In some embodiments, the motion correction computing device 502 is a computing device separate from the workstation 12 and receives MR images acquired by the workstation 12 via a portable storage device such as a flash drive or thumb drive.

[0047] Figure 5BThis is a flowchart of an exemplary method 550 for correcting motion in an image acquired via a PROPELLER pulse sequence. Method 550 can be implemented on a motion correction computing device 502. Method 550 includes receiving k-space data of a three-dimensional (3D) imaging volume of a subject 552, the k-space data of which corresponds to a pre-corrected MR image and is acquired by an MR system by rotating blades in k-space, each blade including multiple views. The k-space data can be acquired via a PROPELLER sequence in a slicing or two-dimensional (2D) manner, wherein the 3D imaging volume includes multiple slices selected by z-gradients, and the kx-ky plane corresponding to the slice is sampled in a PROPELLER manner (i.e., by rotating blades in the kx-ky plane of the slice). Alternatively, the k-space data can be acquired via a PROPELLER sequence in multiple slice encoding steps or in a 3D manner. For example, the 3D k-space (kx-ky-kz) is sampled via a stack of blade sampling along the kx-ky plane in the kz direction. kz corresponds to the slice encoding step. In other words, k-space data is acquired by rotating the blade in the kx-ky plane corresponding to the slice encoding step. The final image is derived by applying Fourier transforms in the kz dimension as well as the kx and ky dimensions. Fourier transforms can be performed on the 3D k-space data in the kz direction to derive k-space data for multiple slices, which are processed similarly to k-space data acquired in a 2D manner.

[0048] In an exemplary embodiment, method 550 further includes receiving 3D MR calibration data of a 3D calibration volume of 554. The calibration data is obtained in a separate acquisition. For example, calibration scans are typically used in MR scans to calculate coil sensitivity maps and other scan parameters for multiple coils. Therefore, the systems and methods described herein do not require additional scans or increase the total scan time. The calibration data can be k-space data, images, or a combination thereof. The 3D imaging volume is equal to or smaller than the 3D calibration volume. Furthermore, method 550 includes jointly estimating rotation and translation based on the k-space data and calibration data of the 3D imaging volume. Unlike method 400, which estimates and corrects rotation before estimating and correcting translation, in this exemplary embodiment, rotation and translation are jointly estimated and corrected in the same operation. Method 550 also includes 558 correcting motion in the pre-corrected image based on the estimated rotation and the estimated translation. Furthermore, method 550 includes outputting a motion-corrected image of 560.

[0049] Figure 6This is a flowchart of an exemplary implementation of jointly estimating 556 rotations and translations. The k-space data acquired in the slices is used only as an illustrative example. The system and method can be applied to k-space data acquired in multiple slice encoding steps. In an exemplary implementation, for each of the multiple slices, rotations and translations are jointly estimated for each blade in the slice using 602. The estimation is performed in image space. In some implementations, rotations and translations are estimated via image registration. For example, to estimate motion in image I, a function of the difference between the transformed image of image I and the target / template image is optimized, with motion parameters such as rotation and translation as optimization variables. The transformed image is image I transformed with motion parameters such as rotation and translation. In PROPELLER, k-space data in the central region R is sampled through all blades 202 (see Figure 2). The central region R is within a circle of diameter L / FOV, where L is the number of views 206 in one blade 202. For each blade, the k-space data in the central region R can be referred to as the central k-space data of the blade and is denoted as S. n It can be based on blade S n The center k-space data of blade 202 is used to reconstruct the center image of the blade. As used in this paper, the center image refers to the image reconstructed from the center k-space data. n The center k-space data is used for blade 202, or for the slice when using the center k-space data of all blades 202 in the slice. In one example, the center k-space data of the blades is meshed onto a Cartesian grid, and then a Fourier transform is performed on the meshed center k-space data of the blades to derive the center image of the blades. When estimating blade rotation and blade translation, the template image is the center image of the slice, which is reconstructed based on the center k-space data of the slice. In one example, the center k-space data of the slice is derived by the following method: for each blade S n The center k-space data is gridded into a Cartesian grid, combining the gridded center k-space data of the blades in the slice, and averaging the combined k-space data for the number of blades 202. The center k-space image of the slice is reconstructed using methods such as performing a Fourier transform on the center k-space data of the slice. Therefore, the blade rotation and blade translation of each blade are estimated based on the center image of the blades and the center image of the slice.

[0050] In the exemplary implementation, the translation of each blade is then fine-tuned, because when jointly estimating the rotation and translation of 602, the translation is based on the k-space data S from the center of the blade. nThe low-resolution image reconstructed from the center k-space data of the slices is estimated and is subject to limitations of the interpolation method. During fine-tuning 604, the translation is also estimated by optimizing the difference function between the transformed image and the template image. Unlike joint estimation 602, in fine-tuning 604, the optimization is repeated, and during each repetition, the template image is updated using the estimated translation from previous iterations; that is, the template image is corrected for translation using the translation estimated in previous iterations. Therefore, the estimated translation has increased accuracy compared to the translation estimated without fine-tuning 604.

[0051] In an exemplary embodiment, the k-space data is motion-corrected once rotation and translation have been estimated for all blades 202. For example, for each blade, the coordinates of the blade's k-space data are rotated with the estimated blade rotation, and a linear phase proportional to the estimated blade translation is removed from the blade's k-space data. In some embodiments, fine-tuning 604 can be skipped, where motion correction uses the rotation and translation estimated in joint estimation 602. The motion-corrected k-space data at the central region R of the blades in the slice is combined 606 to reconstruct a low-resolution central image of the slice. In some embodiments, the k-space data is uncorrected and the central k-space data S n The process 606 is combined to derive a low-resolution center image of the slice. First, joint estimation 602 and fine-tuning 604 are repeated for all slices within the slice, and then this operation is performed for all slices within the 3D imaging volume. Combination 606 is repeated for all slices within the 3D imaging volume. Therefore, multiple center images of the 3D image volume are derived.

[0052] In the depicted implementation, the calibration data can be volumetrically reformatted 608. If the calibration data is k-space data or a mixture of k-space data and an image, the calibration data or k-space data of the calibration data is reconstructed into an image of multiple slices in a 3D calibration volume. The 3D calibration volume may be larger than the 3D imaging volume, or the calibration scan may have a different number and orientation of slices. The calibration data is re-sliced ​​and reoriented such that the slices in the 3D calibration volume match the slices in the 3D imaging volume. That is, each slice in the 3D calibration volume has the same anatomical location as its corresponding slice in the 3D imaging volume. If the 3D calibration volume and the 3D imaging volume are the same and the slices in the calibration and imaging volumes have already matched, reformatting 608 can be skipped. Contrast normalization 610 can be performed on the reformatted 3D calibration data. Since the calibration data was acquired from the k-space data of the 3D imaging volume through different pulse sequences, matrix sizes, resolutions, and scan parameters, the contrast of the calibration data may differ from that of the image acquired via the PROPELLER sequence. Contrast normalization 610 is used to transform the calibration image to have the same or similar contrast as the image in the 3D imaging volume. In some implementations, the contrast of the image of the 3D imaging volume acquired via the PROPELLER sequence is also normalized, and the contrast of the normalized calibration image is the same as or similar to the contrast of the normalized image in the 3D imaging volume.

[0053] In an exemplary embodiment, slice rotation and slice translation are jointly estimated 612 based on a reformatted and / or normalized calibration image and a center image of the slice derived after combining the center k-space data of the 606 blade. In some embodiments, the contrast in the image of the 3D imaging volume is also normalized, and the joint estimation 612 is based on the reformatted and / or normalized calibration image and the normalized center image of the slice. Acquiring calibration data is much faster than acquiring PROPELLER k-space data. For example, for the same 3D volume, calibration data takes approximately 2–6 seconds, while PROPELLER k-space data takes several minutes. Therefore, calibration data contains much less motion than PROPELLER k-space data and can serve as a better motion estimation reference than the center image, thereby increasing the accuracy of motion estimation. Furthermore, since the slices in the PROPELLER k-space data are registered as a joint reference to the calibration data, the motion estimation is consistent across slices compared to method 400. Slice rotation and slice translation can be estimated by using a slice calibration image as a template image, and registering the slice center image to the corresponding slice calibration image by optimizing a function of the difference between the transformed image of the slice center image and the slice calibration image. Slice rotation and slice translation are estimated slice by slice. That is, the joint estimation of slice rotation and slice translation is repeated for multiple slices in the 3D imaging volume. In some implementations, translation is fine-tuned (614). Similar to fine-tuning slice translation (604), fine-tuning slice translation (614) is performed iteratively. The optimization process of estimating slice translation is repeated using a template image, which is updated with a translation-corrected template image based on the translation estimated in previous iterations. Since the resolution of the center image is based only on center k-space data, the resolution of the center image is low. Fine-tuning slice translation (614) improves the estimation accuracy of slice translation. In some implementations, fine-tuning slice translation (614) is skipped. In some implementations, through-plane motion can be estimated by registering slices in the 3D imaging volume to a 3D calibration volume. The estimated through-plane motion can be combined with blade rotation and translation as well as slice rotation and translation to correct motion in PROPELLER images.

[0054] In an exemplary implementation, blade rotation and translation, as well as slice rotation and translation, are combined 616 to derive a final blade rotation and a final blade translation for each blade in the 3D imaging volume. For example, the final rotation is the sum of the blade rotation and the slice rotation, and the final translation is calculated as: Final translation = Blade translation * exp(i * (slice rotation)) + Slice translation.

[0055] In an exemplary implementation, once the final blade rotation and final blade translation are estimated, the estimated final blade rotation and final blade translation are used to correct the 558 motion. For example, the k-space data of each blade 202 is rotationally corrected using the estimated final blade rotation, where the k-space coordinates (kx, ky) are multiplied by a rotation matrix. Where θ represents rotation. The k-space data for each blade 202 is also translated using the estimated final blade translation, where a linear phase proportional to the translation is removed from the blade's k-space data. Unlike motion correction in the combined 606 center k-space data, in motion correction 558, correction is performed on the entire blade or the entire k-space data of the blade. Since the final blade rotation and translation correspond to each blade in each slice, each slice undergoes independent motion correction, allowing for fast reconstruction using parallel processing, where the image or k-space data of the slice or slice-encoded orientation is split into two or more traversals that are processed individually and in parallel with each other.

[0056] Figure 7 This is a comparison of images 702 and 704, which have undergone motion correction using the known method 400, and images 706 and 708, which have undergone motion correction using the method and system described herein. Images 702 and 706 are coronal images based on slices of axial images. Images 704 and 708 are sagittal images based on slices of axial images. Images 702-708 are based on the same k-space data. Compared to images 702 and 704, which have undergone motion correction using the known method 400, images 706 and 708, which have undergone motion correction using the method and system described herein, show a significant reduction in the jagged pattern 424. That is, in method 550, the slices are aligned with each other.

[0057] Figure 8 This is a further comparison of images 801 and 803, which have undergone motion correction using known method 400, and images 805 and 807, which have undergone motion correction using method 550. Images 801 and 805 are axial images of slices near the top of the head, acquired using a PROPELLER pulse sequence, with views 206 in each blade 202 acquired sequentially. Compared to image 805, image 801 has more pronounced crosshairs and ripple artifacts and is less sharp. Images 803 and 807 are images of the moving phantom, acquired using a PROPELLER pulse sequence, with views 206 in each blade 202 acquired in a centered manner. The image quality of image 803 is degraded to the point that image 803 is unusable, indicating that known method 400 fails in motion correction.

[0058] Referring back to Figure 2, each blade 202 includes multiple views 206. When acquiring the blade 202, the views 206 can be acquired sequentially, either starting from the bottom view 206-b and moving upwards, or starting from the top view 206-t and moving downwards. Views 206 can also be acquired in a central manner, where views are acquired starting from the center view 206-c, which is a view that passes through the center 208 of the k-space or is relatively shorter in distance from the center 208 compared to other views 206 and moves outwards. As used herein, the distance from a point (e.g., center 208) to a line (e.g., view 206) is the perpendicular distance from the point to the line. Views 206 farther from the center 208 can be referred to as peripheral views 206-p. For example, if the blade 202 includes eight views as 206-1 to 206-8, where 206-4 is the center view 206-c. In the center-view approach, 206-4 is acquired first, followed by 206-5 or 206-3, 206-6 or 206-2, 206-7 or 206-1, and then 206-8. Since the center view 206-c is closer to the center 208 in k-space, its signal strength is higher than that of the peripheral view 206-p. Due to the low signal-to-noise ratio (SNR) in diffusion-weighted imaging (DWI), especially for high b-values, the view is acquired in the center-view approach to maximize the signal acquired by the DW PROPELLER for each blade 202 via DWI, thus taking advantage of the higher signal strength of the center view 206-c compared to the peripheral view 206-p. However, the peripheral view 206-p is located at a k-space location with a higher spatial frequency and has more detailed spatial location information than the center view 206-c. Therefore, the signal from the peripheral view 206-p is needed for motion correction. Thus, the increased SNR comes at the cost of the signal from the k-space location with the higher spatial frequency. Since known methods 400 use k-space data to estimate motion, they heavily rely on the signal modulation of the k-space data and are unable to estimate motion due to the reduction of signals from higher spatial frequencies during mid-field acquisition. In contrast, the system and method described in this paper are image-based motion estimation that breaks the assumption that signal fluctuations in k-space data originate solely from in-plane motion. Instead, it uses the image structure based on calibration data as a common reference to estimate motion parameters. Therefore, the system and method described in this paper are robust to motion calibration.

[0059] Return to reference Figure 8The image quality of image 803, which undergoes motion correction using the known method 400, is significantly worse than that of image 807, which undergoes motion correction using method 550. The image quality of image 803 is even worse than that of an image without motion correction (not shown). Therefore, the known motion correction method 400 is generally not performed in DWI PROPELLER. In contrast, method 550 is suitable for sequential or center-view images (see images 803, 807). Therefore, method 550 can be used for motion correction of DWI PROPELLER images. In motion correction of DW images, motion correction is performed separately for each diffusion-weighted direction and for the unweighted image (T2 image). For example, if DW images and T2 images for each slice are acquired along the x, y, and z diffusion directions, method 550 is performed separately for the x, y, and z diffusion directions and for the T2 image before merging images from different diffusion directions and before the DW and T2 images are used for further processing and analysis, such as calculating the apparent diffusion coefficient (ADC) mapping. In other words, method 550 is repeated for each diffusion direction and T2 image. If multiple b values ​​are used in the diffusion weighting, method 550 can be repeated for each b value.

[0060] The workstation 12 and motion correction computing device 502 described herein can be any suitable computing device 800 and the software implemented therein. Figure 9 This is a block diagram of an exemplary computing device 800. In an exemplary embodiment, the computing device 800 includes a user interface 804 that receives at least one input from a user. The user interface 804 may include a keyboard 806 that enables the user to input relevant information. The user interface 804 may also include, for example, a pointing device, a mouse, a stylus, a touch-sensitive panel (e.g., a touchpad and a touchscreen), a gyroscope, an accelerometer, a position detector, and / or an audio input interface (e.g., including a microphone).

[0061] Furthermore, in an exemplary embodiment, computing device 800 includes a display interface 817 that presents information (such as input events and / or verification results) to a user. Display interface 817 may also include a display adapter 808 coupled to at least one display device 810. More specifically, in an exemplary embodiment, display device 810 may be a visual display device, such as a cathode ray tube (CRT), liquid crystal display (LCD), light-emitting diode (LED) display, and / or an "electronic ink" display. Alternatively, display interface 817 may include audio output devices (e.g., audio adapters and / or speakers) and / or a printer.

[0062] The computing device 800 also includes a processor 814 and a memory device 818. The processor 814 is connected to a user interface 804, a display interface 817, and the memory device 818 via a system bus 820. In exemplary embodiments, the processor 814 communicates with a user, such as by prompting the user via the display interface 817 and / or by receiving user input via the user interface 804. The term "processor" generally refers to any programmable system, including systems and microcontrollers, reduced instruction set computers (RISCs), complex instruction set computers (CISCs), application-specific integrated circuits (ASICs), programmable logic circuits (PLCs), and any other circuitry or processor capable of performing the functions described herein. The examples above are merely exemplary and are therefore not intended to limit the definition and / or meaning of the term "processor" in any way.

[0063] In an exemplary embodiment, memory device 818 includes one or more devices that enable information (such as executable instructions and / or other data) to be stored and retrieved. Furthermore, memory device 818 includes one or more computer-readable media, such as, but not limited to, dynamic random access memory (DRAM), static random access memory (SRAM), solid-state drive, and / or hard disk. In an exemplary embodiment, memory device 818 stores, but is not limited to, application source code, application object code, configuration data, additional input events, application state, assertion statements, verification results, and / or any other type of data. In an exemplary embodiment, computing device 800 may also include a communication interface 830 coupled to processor 814 via system bus 820. Furthermore, communication interface 830 is communicatively coupled to a data acquisition device.

[0064] In an exemplary embodiment, processor 814 can be programmed by encoding operations using one or more executable instructions and by providing executable instructions in memory device 818. In an exemplary embodiment, processor 814 is programmed to select multiple measurement results received from a data acquisition device.

[0065] In operation, a computer executes computer-executable instructions embodied in one or more computer-executable components stored on one or more computer-readable media to implement aspects of the invention described and / or illustrated herein. Unless otherwise specified, the order of execution or implementation of the operations in the embodiments of the invention shown and described herein is not required. That is, unless otherwise specified, these operations can be performed in any order, and embodiments of the invention may include more or fewer operations than those disclosed herein. For example, it is contemplated that a particular operation may be performed or implemented before, simultaneously with, or after another operation within the scope of various aspects of the invention.

[0066] At least one technical effect of the systems and methods described herein includes (a) motion correction of PROPELLER images with increased accuracy; (b) robust motion correction for all slices or slice coding steps and anatomical structures outside the brain; (c) image-based motion correction; (d) joint estimation of rotation and translation; (e) fine-tuning of translation estimation; (f) motion correction applicable to diffusion-weighted PROPELLER; (g) consistent alignment across slices after motion correction by alignment with calibration data without adversely affecting scan time; and (h) independent motion correction for each slice, thereby allowing for rapid reconstruction using parallel processing.

[0067] Exemplary embodiments of motion correction systems and methods have been described in detail above. These systems and methods are not limited to the specific embodiments described herein, but rather the components of the systems and / or the operation of the methods can be used independently and separately from other components and / or operations described herein. Furthermore, the described components and / or operations may also be defined in other systems, methods, and / or devices, or used in combination with other systems, methods, and / or devices, and are not limited to practice using only the systems described herein.

[0068] Although certain features of various embodiments of the invention may be shown in some figures but not others, this is only for convenience. Any feature of the figures may be referenced and / or claimed in conjunction with any feature of any other figure according to the principles of the invention.

[0069] This written description uses examples to disclose the invention, including the best mode, and also enables those skilled in the art to practice the invention, including making and using any device or system and performing any included methods. The scope of the invention is defined by the claims and may include other examples that would occur to those skilled in the art. Such other examples are intended to fall within the scope of the claims if they have structural elements that are not indistinguishable from the literal language of the claims, or if they include equivalent structural elements that have minor differences from the literal language of the claims.

Claims

1. An MR imaging method for correcting motion in a pre-corrected MR image of a subject, the method comprising: A pulse sequence is applied via an MR system, the blade having a k-space trajectory rotating in k-space, the blade comprising multiple views; The k-space data of the three-dimensional imaging volume of the subject is acquired, and the k-space data of the three-dimensional imaging volume corresponds to the pre-corrected MR image and is acquired through the pulse sequence; Receive three-dimensional MR calibration data of a three-dimensional calibration volume, wherein the three-dimensional calibration volume is greater than or equal to the three-dimensional imaging volume, wherein the three-dimensional MR calibration data is obtained in a separate acquisition from the acquisition of the three-dimensional imaging volume; Based on the k-space data of the three-dimensional imaging volume and the three-dimensional MR calibration data, rotation and translation in the pre-corrected MR image are jointly estimated; Motion in the pre-corrected MR image is corrected based on the estimated rotation and the estimated translation; and Output motion-corrected image.

2. The method of claim 1, wherein the joint estimation of rotation and translation further comprises: The slices in the three-dimensional calibration volume are reformatted to match the slices in the three-dimensional imaging volume; as well as The rotation and translation are jointly estimated based on the k-space data of the three-dimensional imaging volume and the reformulated calibration data.

3. The method of claim 1, wherein the joint estimation of rotation and translation further comprises: Normalize the contrast in the image of the calibration data; as well as The rotation and translation are jointly estimated based on the k-space data of the three-dimensional imaging volume and the normalized calibration data.

4. The method of claim 1, wherein the joint estimation of rotation and translation further comprises: Based on the k-space data of the three-dimensional imaging volume, the blade rotation and blade translation in each blade are jointly estimated.

5. The method of claim 4, wherein the joint estimation of rotation and translation further comprises: The blade translation is iteratively fine-tuned.

6. The method according to claim 1, wherein: Acquiring k-space data also includes acquiring k-space data of multiple slices in the three-dimensional imaging volume, wherein for each of the multiple slices, the k-space data of the three-dimensional imaging volume is acquired by rotating a blade in the kx-ky plane corresponding to the slice; as well as The joint estimation of rotation and translation also includes: For each slice in the three-dimensional imaging volume Combine the k-space data in the central region of the blade of the slice into the central k-space data of the slice; and Reconstruct the central image of the slice based on the central k-space data of the slice; and Slice rotation and slice translation are jointly estimated based on the center images of the multiple slices and the calibration data.

7. The method of claim 6, wherein the joint estimation of rotation and translation further comprises: For each slice in the three-dimensional imaging volume For each blade in the slice Reconstruct the center image of the blade using the center k-space data of the blade; Based on the center image of the blade and the center image of the slice, the blade rotation and blade translation are jointly estimated; as well as The k-space data of the blade is corrected by using the blade rotation and the blade translation; as well as The center image of the slice is updated based on the k-space data of the motion correction of the blade in the slice; as well as Slice rotation and slice translation are jointly estimated based on the updated center image of the multiple slices and the calibration data.

8. The method according to claim 1, wherein: The joint estimation of rotation and translation also includes: Based on the k-space data of the three-dimensional imaging volume, the blade rotation and blade translation in each blade are jointly estimated; Based on the k-space data and calibration data of the three-dimensional imaging volume, slice rotation and slice translation are jointly estimated in each slice of the three-dimensional imaging volume; and The final blade rotation and final blade translation are calculated based on the estimated blade rotation, estimated blade translation, estimated slice rotation, and estimated slice translation; and Correcting motion in the pre-corrected MR image also includes using the final blade rotation and the final blade translation to correct motion in the pre-corrected MR image.

9. The method according to claim 1, wherein acquiring k-space data further comprises acquiring k-space data of the three-dimensional imaging volume in a plurality of slice encoding steps, wherein for each of the plurality of slice encoding steps, the k-space data of the three-dimensional imaging volume is acquired by rotating a blade in the kx-ky plane corresponding to the slice encoding step.

10. The method of claim 1, wherein acquiring k-space data further comprises acquiring k-space data of the three-dimensional imaging volume along multiple diffusion directions; and the method further comprises: For each diffusion weighting direction Rotation and translation are jointly estimated based on the k-space data of the three-dimensional imaging volume according to the diffusion-weighted direction and the calibration data. The motion in the pre-corrected MR image is corrected based on the estimated rotation and the estimated translation, according to the diffusion-weighted direction. as well as Output the motion-corrected image in the diffusion-weighted direction.

11. The method according to claim 10, wherein acquiring k-space data further comprises: The k-space data of the three-dimensional imaging volume is acquired, wherein the k-space data of the three-dimensional imaging volume is acquired by obtaining a view in the blade in a central position.

12. The method according to claim 10, wherein acquiring k-space data further comprises: The k-space data of the three-dimensional imaging volume is acquired by sequentially obtaining views in the blade.

13. The method according to claim 1, wherein acquiring k-space data further includes: The k-space data of the three-dimensional imaging volume is acquired, and the three-dimensional imaging volume is offset from the isocenter of the MR system.

14. The method of claim 1, wherein the joint estimation of rotation and translation further comprises jointly estimating the rotation and translation in the pre-corrected MR image based on an image derived from the k-space data of the three-dimensional imaging volume and an image derived from the calibration data.

15. A motion correction system for correcting motion in pre-corrected MR images of a subject, the motion correction system comprising a motion correction computing device, the motion correction computing device including at least one processor communicating with at least one memory device, and the at least one processor being programmed to: The k-space data of the three-dimensional imaging volume of the subject is received, the k-space data of the three-dimensional imaging volume corresponds to the pre-corrected MR image and is acquired by a pulse sequence having a k-space trajectory of a blade rotating in k-space, the blade including multiple views; Receive three-dimensional MR calibration data of a three-dimensional calibration volume, wherein the three-dimensional calibration volume is greater than or equal to the three-dimensional imaging volume, wherein the three-dimensional MR calibration data is obtained in a separate acquisition from the acquisition of the three-dimensional imaging volume; Based on the k-space data of the three-dimensional imaging volume and the calibration data, rotation and translation in the pre-corrected MR image are jointly estimated; Motion in the pre-corrected MR image is corrected based on the estimated rotation and the estimated translation; and Output the motion-corrected image.

Images