System and method for distortion-free magnetic resonance imaging

Abstract

A method for magnetic resonance imaging is provided herein. The method includes acquiring magnetic resonance data from a subject using a magnetic resonance imaging (MRI) system to sample a hybrid space along a zigzag trajectory. The hybrid space comprises a first axis along a temporal dimension and a second axis along a phase-encoding k-space dimension. The data acquisition involves, for each slice of a plurality of slices, consecutively acquiring all shots of a multi-shot acquisition for the slice before acquiring data for a next slice, with each shot using a radio frequency (RF) excitation pulse having a different flip angle. The method further includes reconstructing images from the acquired magnetic resonance data.

Description

MGH 2024-148-02125141.04924SYSTEM AND METHOD FOR DISTORTION-FREE MAGNETIC RESONANCE IMAGING CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Provisional Patent Application Serial No. 63 / 721,769, filed on November 18. 2024, and entitled -SYSTEM AND METHOD FOR DISTORTION-FREE MAGNETIC RESONANCE IMAGING,” which is herein incorporated by reference in its entirety.STATEMENT OF FEDERALLY SPONSORED RESEARCH

[0002] This invention was made with government support under NS128613, MH111419, AT011429, NS123717, EB030006, EB019437 and EB033206 awarded by the National Institutes of Health. The government has certain rights in the invention.BACKGROUND

[0003] Magnetic resonance imaging (MRI) provides detailed anatomical and functional information about internal body structures. In functional MRI (fMRI) applications, echo-planar imaging (EPI) techniques are commonly employed due to their rapid acquisition capabilities and temporal stability. Single-shot EPI methods offer high stability against physiological fluctuations, but suffer from geometric distortions and T2 / T2* blurring artifacts, which are particularly limiting when high spatial resolutions are desired. These limitations pose challenges for applications requiring submillimeter resolution imaging, such as studies of cerebral cortical columns and layers and small subnuclei in the brain.

[0004] To achieve higher spatial resolution while maintaining reasonable acquisition times, multi-shot imaging techniques have been developed that divide the data acquisition across multiple excitations or “shots.” However, multi-shot approaches introduce vulnerability to physiological processes occurring between shots, including patient motion, cardiac pulsation, respiratory cycles, and other temporal variations. These intershot phase variations can manifest as artifacts that change dynamically over time, introducing temporal instabilities into reconstructed image time series and reducing the temporal signal-to-noise ratio. While various correction methods have been proposed to address these phase variations, they often provide incomplete correction and may not fully restore the temporal stability needed for high-quality functional imaging applications.1QBM25141.04924\99501997.1MGH 2024-148-02125141.04924SUMMARY OF THE DISCLOSURE

[0005] According to an aspect of the present disclosure, a method for magnetic resonance imaging is provided. The method includes acquiring magnetic resonance data from a subject using a magnetic resonance imaging (MRI) system to sample a hybrid space along a zigzag trajectory. The hybrid space includes a first axis along a temporal dimension and a second axis along a phase-encoding k-space dimension. The data acquisition involves, for each slice of a plurality of slices, consecutively acquiring all shots of a multi-shot acquisition for the slice before acquiring data for a next slice, with each shot using a radio frequency (RF) excitation pulse having a different flip angle. The method further includes reconstructing images from the acquired magnetic resonance data.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] FIG. 1 A illustrates an acquisition ordering diagram for conventional multi-shot EPTI.

[0007] FIG. IB illustrates an acquisition ordering diagram for VFA-FLEET EPTI, according to aspects of the present disclosure.

[0008] FIG. 1C illustrates a pulse sequence diagram for VFA-FLEET EPTI, according to aspects of the present disclosure.

[0009] FIG. ID illustrates an encoding pattern in ky-t space for the VFA-FLEET EPTI of FIG. 1C, according to aspects of the present disclosure.

[0010] FIG. 2 illustrates a flowchart of a method for magnetic resonance imaging using VFA-FLEET EPTI.

[0011] FIG. 3 is a block diagram of an example MRI system.DETAILED DESCRIPTION

[0012] Described here are systems and methods for robust high-resolution distortion-free MRI that combine echo-planar time-resolved imaging (EPTI) with Variable-Flip- Angle Fast Low-Excitation-angle Echo-planar Technique (VFA-FLEET) to overcome temporal stabi 1 ity limitations in multi-shot acquisitions. The disclosed systems and methods enable high-resolution, distortion-free, and blurring-free MRI through highly-accelerated multi-shot acquisitions and subspace reconstruction techniques while addressing phase variations that may occur between acquisition segments due to physiological processes or subject motion. The2QBM25141.04924\99501997.1MGH 2024-148-02125141.04924integration of VFA-FLEET may improve temporal stability through acquisition reordering approaches that acquire all excitations of each slice consecutively, and through recursive radiofrequency (RF) pulse design algorithms that may achieve consistent slice profiles and maximized signal intensities across different excitations. These combined techniques may enable efficient acquisition of high-quality' magnetic resonance images suitable for functional imaging, structural imaging, perfusion imaging, and body imaging applications while maintaining the rapid acquisition capabilities and temporal stability needed for clinical and research applications.

[0013] The disclosed systems and methods may be implemented across multiple imaging applications to address diverse clinical and research requirements. Functional imaging applications may utilize the temporal stability improvements provided by the VFA-FLEET acquisition to enhance detection sensitivity7for blood-oxygenation-level-dependent (BOLD) signal changes. The reduced physiological instabilities may enable more reliable detection of neural activation patterns in high-resolution functional MRI studies, particularly at submillimeter spatial resolutions where conventional multi-shot techniques may suffer from motion-related artifacts.

[0014] Structural imaging applications may benefit from the distortion-free and blurring-free image quality7provided by the EPTI acquisition combined with the temporal stability of the VFA-FLEET approach. The method may enable high-resolution anatomical imaging with improved consistency across repeated acquisitions, making the technique suitable for longitudinal studies or applications requiring precise anatomical measurements. The recursive RF pulse design may contribute to consistent signal intensities across different shots, reducing artifacts that could compromise structural image quality.

[0015] Perfusion imaging applications may leverage the multi-echo capabilities of EPTI to provide quantitative perfusion measurements while maintaining temporal stability through the VFA-FLEET acquisition scheme. The method may enable simultaneous acquisition of multiple contrast weightings that can be used to derive perfusion parameters such as cerebral blood flow and cerebral blood volume. The reduced intershot phase variations may improve the reliability of perfusion quantification by minimizing temporal instabilities that could affect the accuracy of perfusion measurements.

[0016] Body imaging applications may utilize the disclosed systems and methods to address motion-related challenges commonly encountered in abdominal or cardiac imaging. The consecutive acquisition of all shots for each slice may reduce sensitivity to respiratory3QBM25141.04924\99501997.1MGH 2024-148-02125141.04924motion and cardiac pulsation compared to conventional multi-shot approaches where longer time delays between shots can introduce motion artifacts. The method may be particularly beneficial for body imaging applications requiring high spatial resolution or quantitative measurements.

[0017] EPTI is an MRI technique that addresses limitations inherent in conventional EPI. While EPI may be commonly used for fMRI due to fast acquisition times and high stability, EPI typically produces images with lower resolution and with distortion artifacts. EPTI may provide high-resolution, distortion-free, and blurring-free magnetic resonance images through highly -accelerated multi-shot acquisitions and subspace reconstruction techniques. Multi-shot EPTI may require multiple excitations or “shots” to achieve high resolution imaging. In conventional multi-shot EPTI implementations, time delays between different shots of a single slice can introduce signal variations due to patient motion or other physiological processes. These phase variations induced by physiological processes or motion occurring between shots may introduce artifacts that change with time causing instabilities in reconstructed time-series data, which cannot be fully corrected with either navigator-free or navigator-based correction methods.

[0018] EPTI implements a sampling strategy that addresses the geometric distortions and signal blurring commonly associated with conventional EPI techniques. The EPTI approach samples a hybrid space that spans both the phase-encoding dimension and the temporal dimension, which may be referred to as ky-t space, where kyrepresents the phaseencoding k-space dimension and t represents the temporal dimension.

[0019] By way of example, the EPTI acquisition scheme may utilize a zigzag traj ectory pattern to sample the hybrid ky-t space. This zigzag pattern may include multiple linear sections or segments, where each segment contains a plurality of temporally adjacent data samples that are spaced apart along the phase-encoding dimension. The data samples within each linear section may be acquired in a temporally sequential manner, with temporally adjacent samples separated by a first temporal spacing that may be on the order of milliseconds. This short temporal spacing may minimize Bo-inhomogeneity induced phase variations and T2 or T2* decay effects that can contribute to image artifacts. The zigzag trajectory may implement an interleaved acceleration approach in the phase-encoding direction. Data samples on adjacent linear sections of the sampling pattern may be interleaved in the phase-encoding dimension, such that neighboring kypoints are acquired with minimal temporal separation. This interleaving strategy may ensure that phase-encoding lines that are adjacent in k-space are4QBM25141.04924\99501997.1MGH 2024-148-02125141.04924separated by a larger temporal spacing than the first temporal spacing between temporally adjacent samples within each linear section.

[0020] EPTI acquisitions may be performed using multiple shots or segments to achieve complete coverage of the desired ky-t space. In some implementations, phase-encoding lines covering the positive half of the kyaxis may be sampled in a first shot while phaseencoding lines covering the negative half of the kyaxis may be sampled in a second shot. The hybrid space may alternatively be partitioned into additional segments to enable greater acceleration of data acquisition, with each shot covering a different portion of the ky-t space.

[0021] The EPTI technique may provide temporal signal evolution information across the echo-planar imaging readout window. This temporal information may enable the reconstruction of images with temporal resolution that may approach the echo spacing interval. The method may generate dozens or hundreds of images across the EPI readout window, spaced at time intervals that may be approximately equal to the echo-time spacing.

[0022] Image reconstruction from EPTI data may utilize specialized algorithms that can handle the undersampled ky-t space data. Reconstruction approaches may include tilted kernel-based methods that interpolate undersampled data to synthesize additional data points in the hybrid space, subspace reconstructions, and so on. These reconstruction kernels may be oriented at specific angles with respect to the phase-encoding dimension and may span multiple points in both the phase-encoding and temporal dimensions. The reconstruction process may also incorporate coil sensitivity information from multi-channel receiver coils to improve the conditioning and signal -to-noise ratio of the reconstructed images.

[0023] EPTI may be implemented with various pulse sequence configurations, including gradient-echo sequences, spin-echo sequences, or combined gradient-echo and spinecho approaches. The technique may be extended to three-dimensional acquisitions by incorporating partition-encoding dimensions, creating a ky-kz-t hybrid space for volumetric imaging applications. Simultaneous multi-slice techniques may also be integrated with EPTI to further accelerate data acquisition by exciting and acquiring data from multiple slices simultaneously.

[0024] The EPTI approach may enable fMRI and quantitative parameter mapping applications, including T2 relaxometry, T2 mapping, and susceptibility -weighted imaging. The high temporal resolution capabilities may provide improved accuracy for functional and / or quantitative measurements compared to conventional imaging approaches. The distortion-free and blurring-free characteristics of EPTI may make the technique particularly suitable for5QBM25141.04924\99501997.1MGH 2024-148-02125141.04924applications requiring high spatial fidelity, such as fMRI, diffusion imaging, and perfusion imaging studies.

[0025] The temporal stability challenges in conventional multi-shot EPTI may be addressed through integration with VFA-FLEET. In these instances, VFA-FLEET can improve the stability of multi-shot acquisitions through two primary approaches: acquisition reordering and recursive RF pulse design.

[0026] The acquisition reordering implemented in VFA-FLEET may implement a shot-sequential pattern where all shots for a given slice are acquired consecutively before the system moves to acquire data from the next slice. In this reordered approach, the MRI system may complete the entire multi-shot acquisition sequence for slice one, then proceed to complete the entire multi-shot acquisition sequence for slice two, and continue this pattern through all slices in the imaging volume. This reordering may eliminate the temporal delays that occur between shots of the same slice in conventional EPTI acquisitions.

[0027] As described above, the temporal delays present in conventional EPTI may create opportunities for phase variations to develop between different shots of the same slice. Patient motion, respirator}' cycles, cardiac pulsation, and other physiological processes may introduce phase changes during the time intervals between shots. These phase variations may manifest as inconsistencies in the magnetic resonance signal that can degrade image quality and temporal stability in the reconstructed time-series data. The consecutive acquisition of all shots for each slice in the VFA-FLEET reordering scheme minimizes the time window during which phase variations can develop between shots. By acquiring all shots for a given slice in rapid succession, the disclosed systems and methods may reduce exposure to motion-induced phase changes and physiological fluctuations that occur over longer time scales. The shortened temporal window between shots may result in more consistent phase relationships across the multiple excitations used to reconstruct each slice.

[0028] The acquisition reordering of VFA-FLEET may provide particular benefits for imaging applications where temporal stability is important for data quality. For instance, fMRI applications may benefit from reduced shot-to-shot phase variations, as these variations can introduce temporal instabilities that may interfere with the detection of activation-related signal changes. The improved temporal stability achieved through acquisition reordering may enhance the reliability of time-series measurements and improve the signal-to-noise characteristics of the acquired data. The reordered acquisition pattern may also influence the overall timing characteristics of the imaging sequence. While conventional EPTI may6QBM25141.04924\99501997.1MGH 2024-148-02125141.04924distribute the shots for each slice across the entire repetition time, the VFA-FLEET reordering may concentrate all shots for each slice within a shorter temporal window.

[0029] The recursive RF pulse design implemented with VFA-FLEET addresses signal intensity inconsistencies between different excitations that can arise from acquisition reordering. These inconsistencies may result from imperfections in RF slice profdes and intrinsic spin relaxation processes. The recursive RF pulse design can achieve consistent slice profiles for different excitations and maximized signal intensities to suppress artifacts and improve signal-to-noise ratio.

[0030] By way of example, the recursive RF pulse design algorithm may utilize Shinnar-LeRoux (SLR) RF pulses to achieve consistent slice profiles across different excitations. The recursive calculation of flip angles may follow a specific mathematical relationship that ensures optimal signal characteristics while maintaining slice profile consistency throughout the multi-shot acquisition sequence. For instance, the flip angles for different shots in the VFA-FLEET acquisition may be recursively calculated as,0^ -tan-1(sin (6> )) (1);

[0031] where i is the shot number index. The last excitation pulse typically has a 90 degree flip angle to maximize signal by depleting the remaining longitudinal magnetization. This recursive relationship may provide a systematic approach to determining the sequence of flip angles that maintains consistent excitation characteristics across multiple shots while accounting for the effects of previous excitations on the magnetization state.

[0032] For 3-shot acquisitions, the flip angle sequence may be set to [35.3°, 45°, 90°]. In this configuration, the final flip angle may be set to 90°, and the preceding flip angles may¬ be calculated using the recursive formula working backwards from this final value. The 35.3° flip angle may be applied to the first shot, followed by 45° for the second shot, and 90° for the final shot of each slice acquisition.

[0033] For 5-shot acquisitions, the flip angle sequence may be set to [26.6°, 30°, 35.3°, 45°, 90°]. Similar to the 3-shot configuration, the final flip angle may be set to 90°, with the preceding four flip angles calculated recursively. The sequence may begin with 26.6° for the first shot, progress through 30°, 35.3°, and 45° for the subsequent shots, and conclude with the 90° flip angle for the final shot of each slice.

[0034] Using standard sine excitation RF pulses scaled to the desired variable flip angles may result in inconsistent slice profiles across shots, which can lead to varying signal levels causing imaging artifacts. A recursive SLR pulse design algorithm can be used to 7QB\125141.04924\99501997.1MGH 2024-148-02125141.04924produce consistent slice profiles across shots. In this scheme, the first RF pulse is designed with the conventional SLR algorithm. The second RF pulse is then calculated based on knowledge of the residual longitudinal magnetization remaining after the first pulse as a function of through-slice position and uses a slice profile target identical to the first pulse. In this recursive manner, all the RF pulses for different shots can be jointly designed.

[0035] The recursive flip angle calculation may optimize signal intensity while compensating for the effects of Ti relaxation and other factors that could otherwise lead to signal inconsistencies between shots. By using progressively increasing flip angles culminating in the 90° final excitation, the algorithm may maximize the available magnetization for signal generation while maintaining the temporal stability benefits of the consecutive shot acquisition approach.

[0036] As a non-limiting example, the RF pulse parameters may include a timebandwidth product of 4-6, passband ripples of 0.01%, stopband ripples of 1%, and pulse durations ranging from 4.0 to 4.4 milliseconds depending on slice thickness. In some cases, the RF pulse design may incorporate Blackman filter windowing outside the central 1.5 x region to optimize pulse characteristics and reduce artifacts.

[0037] Diagrams depicting the acquisition ordering of conventional multi-shot EPTI and VFA-FLEET EPTI are shown in FIGS. 1 A and IB, respectively . In VFA-FLEET, all shots of a given slice are acquired consecutively with minimal delay between them. By way of example, shots may be acquired approximately 80 ms apart. Because of the short time intervals between the multiple shots of encoding used to form one slice in VFA-FLEET, there is negligible time for longitudinal relaxation, and hence it can be ignored. The longitudinal magnetization available before each RF excitation is a function of both the previous excitation pulse’s rotation parameters and the longitudinal magnetization available immediately before the previous excitation pulse.

[0038] FIGS. 1C and ID show an example pulse sequence diagram for VFA-FLEET EPTI and the corresponding encoding pattern in ky-t space, respectively. In the illustrated example of VFA-FLEET EPTI, after each excitation with the chosen RF pulse, scaled to the desired flip angle of the z-th shot (0 / ), a navigator echo can be acquired to monitor intershot phase variations and provide phase correction data. By way of example, the navigator echo may be a three-line one-dimensional navigator echo. The navigator echo acquisition may¬ enable real-time assessment of physiological motion effects during the imaging sequence.8QBM25141.04924\99501997.1MGH 2024-148-02125141.04924

[0039] To suppress unwanted echo refocusing in VFA-FLEET, a spoiling gradient is applied along the slice-select axis after each image echo readout. As a non-limiting example, the spoiling gradient moment may be set to provide adequate suppression while maintaining reasonable acquisition times. This spoiling approach may contribute to the temporal stability improvements achieved with the VFA-FLEET technique.

[0040] In EPTI, readout lines are spaced apart in time by the echo spacing (TeSp) and are spaced apart in the phase-encoding (PE) direction by ATE. with ATE =1 corresponding to the fully sampled case. 7?Segdenotes the coverage along the PE direction of each EPTI shot. The segment coverage parameters (e.g., / tyg may define the coverage along the phase-encoding direction for each EPTI shot. In some cases, Jisegvalues may range from 8 to 16 depending on the total number of shots and desired spatial resolution.

[0041] In EPTI, spatiotemporal controlled aliasing in parallel imaging (CAIPI) sampling can additionally be used by alternating the PE encoding for different readout lines for better use of coil sensitivities across the RF coil array and better use of temporal correlation during image reconstruction. CAIPI may be achieved by phase-shifting the bands of a multiband RF excitation pulse. By way of example, a blipped-CAIPI simultaneous multislice (SMS) method may be incorporated into VFA-FLEET EPTI. As shown in FIG. 1C, gradient blips may be applied along the slice axis during readout to introduce field-of-view shifts in the phase-encoding direction across different slices. These gradient blips may improve the g-factor performance of SMS acquisitions by optimizing the spatial separation of simultaneously excited slices. The blipped-CAIPI SMS integration may enable simultaneous acquisition of multiple slices with reduced g-factor penalties compared to conventional SMS approaches. As non-limiting examples, the FOV shift introduced by the gradient blips may be set to FOV / 2 or FOV / 3 or other values depending on the number of simultaneously acquired slices and the specific coil geometry. In some cases, SMS factors of 2 or 3 or higher may be implemented to balance temporal resolution improvements with acceptable noise amplification levels.

[0042] To train the reconstruction to recover the highly undersampled k-space data acquired in EPTI (FIG. ID), fully sampled (fcc-fc-fc-t) low-resolution calibration data are acquired. In EPTL multiple images, each at different TEs along the echo train, are reconstructed. To maximize the ability to accelerate the data acquisition while maintaining image reconstruction performance, a subspace image-reconstruction algorithm can be implemented. In subspace reconstruction, the signal evolution space (T2* decay for gradientecho EPTI) within specific quantitative parameter value ranges is first simulated based on the9QBM25141.04924\99501997.1MGH 2024-148-02125141.04924Bloch equations and the acquisition parameters. Then, a small set of basis vectors (e.g., three basis vectors) is extracted through principal component analysis, denoted as ( / ) . These vectors form a low-dimensional subspace that can closely approximate the entire signal space of interest. Using this basis, the image series across the readout echo train can be calculated by ( / >c , where c is the coefficient maps of the basis vectors. In this way, the degrees of freedom of the reconstruction are reduced from the number of echoes to the number of basis vectors, improving the conditioning of reconstruction and thus the SNR of images.

[0043] By way of example, the subspace reconstruction can computed by,

[0044] where B is the phase evolution across different image echoes due to Bo inhomogeneity, which can be obtained from a calibration scan; S is the coil sensitivity; is the Fourier transform operator; U is the undersampling mask; and y is the acquired undersampled k-space dataset. The undersampling mask U may define the specific k-space sampling pattern used in the EPTI acquisition, incorporating both the phase-encoding acceleration and any spatiotemporal sampling scheme (e.g., CAIPI, blipped-CAIPI). The mask may account for the alternating phase-encoding patterns across different readout lines and the specific trajectory through k-space-time implemented in the EPTI sequence.

[0045] The regularization term R(c) can be incorporated to further improve the conditioning, and z? is the control parameter of the regularization. The regularization term R(c) in may incorporate spatial and temporal constraints to improve reconstruction quality and reduce noise amplification. Spatial regularization may be applied to the coefficient maps to promote smoothness and reduce artifacts, while temporal regularization may exploit correlations between different echo times to enhance reconstruction stability. As anon-limiting example, a locally low-rank regularization term can be used. After solving for c, images across the echo train can be recovered by the product (f)C .

[0046] The subspace reconstruction algorithm may be implemented to reconstruct multiple images at different echo times along the echo train from the highly undersampled k-space data acquired in EPTI. The reconstruction may be formulated as an optimization problem that minimizes the expression in Eqn. (2). As a non-limiting example, the subspace reconstruction may utilize three basis vectors extracted through principal component analysis to form a low-dimensional subspace that approximates the signal evolution space. The signal evolution space may be simulated based on the Bloch equations and the specific acquisition 10QBM25141.04924\99501997.1MGH 2024-148-02125141.04924parameters for the gradient-echo EPTI sequence. The three basis vectors may capture the dominant signal decay patterns across the echo train, enabling efficient representation of the T2* decay behavior with reduced computational complexity.

[0047] The basis vectors may be derived by simulating signal evolution within specific quantitative parameter value ranges corresponding to typical brain tissue properties. Principal component analysis may be applied to the simulated signal evolution curves to extract the three most significant basis functions. These basis vectors may form a subspace that closely approximates the entire signal space of interest while reducing the dimensionality of the reconstruction problem. As described above, the image series across the readout echo train may be represented as a linear combination of the three basis vectors, where the coefficient maps c determine the contribution of each basis vector at each spatial location. The coefficient maps may be estimated through the optimization process, and the final images at different echo times may be reconstructed by combining the basis vectors with their corresponding coefficient maps.

[0048] For VFA-FLEET EPTI, the signal equation may be simplified to SVFA-FLEET = Mo sin 0i eTI'T2. where 0i represents the first flip angle in the variable flip angle sequence. This simplified expression may result from the assumption of full recovery of longitudinal magnetization before the first excitation due to the long volume repetition time and the consecutive acquisition of all shots for each slice with minimal delays between shots.

[0049] Referring now to FIG. 2. a flowchart is illustrated as setting forth the steps of an example method for imaging a subject using a VFA-FLEET EPTI acquisition. The method 200 may begin at step 202 by positioning the subject within the MRI scanner and performing initial setup procedures, including shimming, calibration scans, and coil sensitivity mapping. The setup may include acquiring Bo field maps and coil sensitivity maps that may be used during subsequent reconstruction processes.

[0050] At step 204, the method may acquire fully sampled low-resolution calibration data in kx-ky-kz-t space to train the subspace reconstruction algorithm. The calibration data may provide reference information for the undersampled EPTI acquisition and may enable estimation of coil sensitivities and Bo inhomogeneity patterns across the imaging volume.

[0051] At step 206, the method may initialize the pulse sequence parameters and generate the recursive RF pulse sequence for the VFA-FLEET acquisition (e.g., using an SLR algorithm). The pulse sequence initialization may include setting the number of slices to be acquired, the number of shots per slice (e.g., 3-shot. configuration, 5-shot configuration, etc.).11QBM25141.04924\99501997.1MGH 2024-148-02125141.04924the spatial resolution parameters (e.g., 0.8 mm, 1 mm, or 2 mm isotropic voxel sizes), the echo spacing (Tesp), the repetition time (TR), and the simultaneous multislice acceleration factor if SMS is to be employed. Additional parameters may include the phase-encoding acceleration factor (RPE), the segment coverage parameter (Rscg) defining the coverage along the phaseencoding direction for each EPTI shot, and the field-of-view dimensions. The RF pulses may be designed to achieve consistent slice profiles across different shots while implementing the variable flip angle scheme. The flip angles may be calculated recursively according to Eqn (1), with the final flip angle typically set to 90°.

[0052] The VFA-FLEET EPTI method may be applied to different spatial resolutions to accommodate various imaging requirements. For high-resolution applications, 0.8 mm isotropic voxel sizes may be achieved using 3-shot acquisitions without SMS acceleration. The 0.8 mm isotropic resolution may provide submillimeter spatial detail for applications such as cortical layer imaging or high-resolution functional studies.

[0053] For balanced resolution and acquisition efficiency, 1 mm isotropic voxel sizes may be implemented using 3-shot or 5-shot configurations. The 1 mm isotropic protocol may incorporate SMS acceleration with factors of 2 to maintain reasonable temporal resolution while achieving whole-brain coverage. This spatial resolution may be suitable for functional MRI studies requiring both high spatial detail and adequate temporal sampling.

[0054] For applications prioritizing temporal resolution or signal-to-noise ratio, 2 mm isotropic voxel sizes may be employed with 3-shot acquisitions. The 2 mm isotropic resolution may enable faster acquisition times and higher temporal signal-to-noise ratios due to increased voxel volume and reduced acceleration requirements. This resolution may be appropriate for functional connectivity studies or applications where temporal dynamics are of primary interest.

[0055] Data are then acquired from the subject, as indicated at process block 208. At step 210, the VFA-FLEET EPTI data acquisition is initiated by selecting the first slice for imaging. Then, at step 212, the multi-shot acquisition for the selected slice is performed using the predetermined flip angle sequence for the initialized pulse sequence. Each shot may utilize the corresponding recursively designed RF pulse, followed by the EPTI readout with spatiotemporal CAIPI encoding if implemented. Navigator echoes may be acquired after each RF excitation to monitor phase variations and provide correction data. In some cases, spoiling gradients may be applied along the slice-select axis after each image echo readout to suppress unwanted echo refocusing.12QBM25141.04924\99501997.1MGH 2024-148-02125141.04924

[0056] As described above, the acquisition reordering scheme ensures that all shots for the selected slice are acquired consecutively before proceeding to subsequent slices, thereby minimizing intershot phase variations due to physiological processes. Therefore, at step 214, a determination is made whether all shots for the current slice have been completed. If additional shots remain for the current slice, the method may return to step 212 to acquire the next shot with the appropriate flip angle. If all shots for the current slice have been acquired, the method may proceed to step 216.

[0057] At step 216, a determination is made whether additional slices have yet to be acquired. If more slices remain in the imaging volume, the method may proceed to step 218 to select the next slice and return to step 212 to begin the multi-shot acquisition for that slice. At step 218, the method may select the next slice in the acquisition sequence and prepare for the multi-shot acquisition. The slice selection may follow a predetermined ordering scheme that may optimize acquisition efficiency and minimize cross-slice interference effects. If all slices have been acquired, the method may proceed to step 220.

[0058] At step 220, the method may apply ghost correction schemes to the acquired k-space data. The correction may include intrashot first-order phase correction for Nyquist ghost correction using the navigator echo data (if acquired), intershot phase correction to align phases across different shots, and / or intershot magnitude normalization to equalize signal levels between shots.

[0059] The ghost correction schemes may process the raw k-space data to address phase and magnitude inconsistencies across shots and within individual shots. The intrashot phase correction may operate first, utilizing the navigator echo data to correct Nyquist ghost artifacts within each shot independently. The intrashot phase correction may calculate first-order phase correction parameters from the navigator echo data by comparing the phase evolution across the navigator lines. The phase correction parameters may be applied to the image data within each shot to align the phases of alternating readout lines and reduce ghosting artifacts. This correction may be performed independently for each shot and each coil element to account for shot-specific and coil-specific phase variations.

[0060] The intershot phase correction may process the phase-corrected data from the intrashot correction to align phases across different shots of the same slice. This correction scheme may address phase variations that occur between shots due to physiological processes, patient motion, or other temporal instabilities during the acquisition. The intershot phase correction may utilize the navigator echo data to estimate phase differences between shots and13QBM25141.04924\99501997.1MGH 2024-148-02125141.04924apply appropriate phase adjustments to align the phases across all shots contributing to a single slice. For instance, the intershot phase correction may calculate phase correction parameters by comparing the navigator echo phases from different shots and determining the phase offsets that minimize phase differences across shots. The phase correction may be applied to the image data from each shot to ensure phase consistency across the multi-shot acquisition. This correction may be particularly beneficial for conventional EPTI acquisitions where longer time delays between shots can introduce significant phase variations.

[0061] The intershot magnitude normalization may operate on the phase-corrected data to equalize signal levels across different shots. This correction scheme may address magnitude variations that can occur between shots due to signal intensity fluctuations, coil sensitivity changes, or other factors affecting signal amplitude consistency. The magnitude normalization may utilize the navigator echo magnitude data to calculate normalization factors that equalize signal levels across shots. By way of example, the intershot magnitude normalization may calculate normalization factors by analyzing the navigator echo magnitudes from different shots and determining scaling factors that minimize the sum-of-square differences between navigator magnitudes. The normalization factors may be applied to the image data from each shot to achieve consistent signal levels across the multi-shot acquisition. This correction may help reduce artifacts related to signal intensity variations and improve the overall stability of the reconstructed time-series data.

[0062] The three ghost correction schemes may be applied sequentially, with each correction addressing specific types of artifacts and inconsistencies. The combination of intrashot phase correction, intershot phase correction, and intershot magnitude normalization may provide comprehensive artifact reduction and improve the temporal stability of the VFA-FLEET EPTI acquisition compared to uncorrected data.

[0063] At step 222, images are reconstructed from the acquired k-space data. By way of example, the method may perform subspace reconstruction of the corrected k-space data to generate multiple images at different echo times. The reconstruction may utilize the three basis vectors extracted through principal component analysis to represent the signal evolution space, and may solve the optimization problem defined in Equation (2) to estimate the coefficient maps. The final reconstructed images may be generated by combining the estimated coefficient maps with the basis vectors. The reconstruction may produce multiple contrast- weighted images corresponding to different echo times along the echo train, providing distortion-free and blurring-free images with improved temporal stability compared to conventional multi-14QBM25141.04924\99501997.1MGH 2024-148-02125141.04924shot EPTI approaches. Alternative reconstruction techniques may include tilted kernel-based methods that interpolate undersampled data to synthesize additional data points in the hybrid space, compressed sensing approaches with sparsity constraints, or machine learning-based reconstruction algorithms.

[0064] At step 224, the method may optionally apply post-processing techniques to the reconstructed images. For example, complex-domain NORDIC denoising may be implemented to enhance temporal signal-to-noise ratio performance by selectively removing thermal noise components while preserving physiological signal variations. The denoised images may provide improved sensitivity7for functional imaging applications or quantitative parameter mapping.

[0065] NORDIC denoising may utilize noise reduction with distribution correction to identify and remove spatially7uncorrelated thermal noise while preserving physiologically relevant signal variations. The complex-domain implementation may operate on both magnitude and phase components of the image data to provide comprehensive noise reduction. By way of example, the NORDIC denoising algorithm may estimate thermal noise levels using Marchenko-Pastur principal component analysis to determine the noise floor and identify noise-dominated principal components. The method may calculate g-factor maps to account for parallel imaging noise amplification and ensure accurate noise estimation across different spatial locations. The denoising process may selectively remove noise components while preserving signal components that contribute to physiological contrast and temporal dynamics.

[0066] The reconstructed images can then be displayed to a user, stored for later use or further processing, or both, as indicated at step 226. Further processing of the reconstructed images may encompass a wide range of analysis techniques depending on the specific imaging application. For functional imaging applications, further processing may include statistical analysis to identify regions of neural activation, connectivity analysis to examine functional networks, or temporal filtering to enhance signal-to-noise characteristics. For structural imaging applications, additional processing may involve tissue segmentation, morphometric analysis, or registration to standard anatomical templates. For perfusion imaging applications, further processing may include pharmacokinetic modeling to derive quantitative perfusion parameters such as cerebral blood flow, cerebral blood volume, and mean transit time from the multi-echo EPTI data.

[0067] The display of reconstructed images may be implemented through various visualization interfaces that enable real-time or near-real-time viewing of the images. The15QBM25141.04924\99501997.1MGH 2024-148-02125141.04924images may be presented as individual echo time images, time-series data for functional imaging applications, or multi-contrast displays that showcase the different tissue contrasts available across the echo train. Interactive display tools may allow users to navigate through different echo times, adjust windowing and leveling parameters, and overlay functional activation maps on anatomical images for fMRI applications.

[0068] Referring particularly now to FIG. 3, an example of an MRI system 300 that can implement the methods described here is illustrated. The MRI system 300 includes an operator workstation 302 that may include a display 304, one or more input devices 306 (e.g., a keyboard, a mouse), and a processor 308. The processor 308 may include a commercially available programmable machine running a commercially available operating system. The operator workstation 302 provides an operator interface that facilitates entering scan parameters into the MRI system 300. The operator workstation 302 may be coupled to different servers, including, for example, a pulse sequence server 310, a data acquisition server 312, a data processing server 314, and a data store server 316. The operator workstation 302 and the servers 310, 312. 314, and 316 may be connected via a communication system 340, which may include wired or wireless network connections.

[0069] The pulse sequence server 310 functions in response to instructions provided by the operator workstation 302 to operate a gradient system 318 and a radiofrequency (RF) system 320. Gradient waveforms for performing a prescribed scan are produced and applied to the gradient system 318, which then excites gradient coils in an assembly 322 to produce the magnetic field gradients Gx, G , and G_ that are used for spatially encoding magnetic resonance signals. The gradient coil assembly 322 forms part of a magnet assembly 324 that includes a polarizing magnet 326 and a whole-body RF coil 328.

[0070] RF waveforms are applied by the RF system 320 to the RF coil 328, or a separate local coil to perform the prescribed magnetic resonance pulse sequence. Responsive magnetic resonance signals detected by the RF coil 328, or a separate local coil, are received by the RF system 320. The responsive magnetic resonance signals may be amplified, demodulated, filtered, and digitized under direction of commands produced by the pulse sequence server 310. The RF system 320 includes an RF transmitter for producing a wide variety' of RF pulses used in MRI pulse sequences. The RF transmitter is responsive to the prescribed scan and direction from the pulse sequence server 310 to produce RF pulses of the desired frequency, phase, and pulse amplitude waveform. The generated RF pulses may be applied to the whole-body RF coil 328 or to one or more local coils or coil arrays.16QBM25141.04924\99501997.1MGH 2024-148-02125141.04924

[0071] The RF system 320 also includes one or more RF receiver channels. An RF receiver channel includes an RF preamplifier that amplifies the magnetic resonance signal received by the coil 328 to which it is connected, and a detector that detects and digitizes the I and Q quadrature components of the received magnetic resonance signal. The magnitude of the received magnetic resonance signal may, therefore, be determined at a sampled point by the square root of the sum of the squares of the I and Q components:M= jl2+Q2;

[0072] and the phase of the received magnetic resonance signal may also be determined according to the following relationship:>

[0073] The pulse sequence server 310 may receive patient data from a physiological acquisition controller 330. By way of example, the physiological acquisition controller 330 may receive signals from a number of different sensors connected to the patient, including electrocardiograph (ECG) signals from electrodes, or respiratory signals from a respiratory bellows or other respiratory’ monitoring devices. These signals may be used by the pulse sequence server 310 to synchronize, or ‘'gate,’’ the performance of the scan with the subject’s heart beat or respiration. The pulse sequence server 310 is operable to adjust acquisition parameters such as the number of shots, spatial resolution, SMS acceleration factors, and echo spacing. The flexible parameter selection may enable the system to balance temporal resolution, spatial resolution, and signal-to-noise ratio requirements for specific imaging applications while maintaining the temporal stability benefits provided by the VFA-FLEET approach.

[0074] The MRI system 310 operation may incorporate feedback mechanisms that monitor acquisition quality and adjust parameters as needed during the imaging session. The navigator echo data may provide real-time feedback about intershot phase variations and signal stability, enabling adaptive adjustments to acquisition parameters if excessive motion or instability’ is detected. The feedback system may help maintain optimal image quality throughout the acquisition process.

[0075] The pulse sequence server 310 may also connect to a scan room interface circuit 332 that receives signals from various sensors associated with the condition of the patient and17QB\125141.04924\99501997.1MGH 2024-148-02125141.04924the magnet system. Through the scan room interface circuit 332, a patient positioning system 334 can receive commands to move the patient to desired positions during the scan.

[0076] The digitized magnetic resonance signal samples produced by the RF system 320 are received by the data acquisition server 312. The data acquisition server 312 operates in response to instructions downloaded from the operator w orkstation 302 to receive the realtime magnetic resonance data and provide buffer storage, so that data is not lost by data overrun. In some scans, the data acquisition server 312 passes the acquired magnetic resonance data to the data processor server 314. In scans that require information derived from acquired magnetic resonance data to control the further performance of the scan, the data acquisition server 312 may be programmed to produce such information and convey it to the pulse sequence server 310. For example, during pre-scans, magnetic resonance data may be acquired and used to calibrate the pulse sequence performed by the pulse sequence server 310. As another example, navigator signals may be acquired and used to adjust the operating parameters of the RF system 320 or the gradient system 318, or to control the view order in w hich k-space is sampled. In still another example, the data acquisition server 312 may also process magnetic resonance signals used to detect the arrival of a contrast agent in a magnetic resonance angiography (MRA) scan. For example, the data acquisition server 312 may acquire magnetic resonance data and processes it in real-time to produce information that is used to control the scan.

[0077] The data processing server 314 receives magnetic resonance data from the data acquisition server 312 and processes the magnetic resonance data in accordance with instructions provided by the operator workstation 302. Such processing may include, for example, reconstructing two-dimensional or three-dimensional images by performing a Fourier transformation of raw k-space data, performing other image reconstruction algorithms (e.g., iterative or backproj ection reconstruction algorithms), applying fdters to raw k-space data or to reconstructed images, generating functional magnetic resonance images, or calculating motion or flow images.

[0078] Images reconstructed by the data processing server 314 are conveyed back to the operator workstation 302 for storage. Real-time images may be stored in a data base memory cache, from which they may be output to operator display 302 or a display 336. Batch mode images or selected real time images may be stored in a host database on disc storage 338. When such images have been reconstructed and transferred to storage, the data processing server 314 may notify the data store server 316 on the operator workstation 302. The operator18QBM25141.04924\99501997.1MGH 2024-148-02125141.04924workstation 302 may be used by an operator to archive the images, produce films, or send the images via a network to other facilities.

[0079] The MRI system 300 may also include one or more networked workstations 342. For example, a networked workstation 342 may include a display 344, one or more input devices 346 (e.g., a keyboard, a mouse), and a processor 348. The networked workstation 342 may be located within the same facility as the operator workstation 302, or in a different facility, such as a different healthcare institution or clinic.

[0080] The networked workstation 342 may gain remote access to the data processing sen' er 314 or data store server 316 via the communication system 340. Accordingly, multiple networked workstations 342 may have access to the data processing server 314 and the data store server 316. In this manner, magnetic resonance data, reconstructed images, or other data may be exchanged between the data processing server 314 or the data store server 316 and the networked workstations 342, such that the data or images may be remotely processed by a networked workstation 342.

[0081] The present disclosure has described one or more preferred embodiments, and it should be appreciated that many equivalents, alternatives, variations, and modifications, aside from those expressly stated, are possible and within the scope of the disclosure.19QBM25141.04924\99501997.1

Claims

MGH 2024-148-02125141.04924CLAIMS1. A method for magnetic resonance imaging, comprising:acquinng magnetic resonance data from a subject using a magnetic resonance imaging (MRI) system to sample a hybrid space along a zigzag trajectory, wherein the hybrid space comprises a first axis along a temporal dimension and a second axis along a phase-encoding k-space dimension, wherein acquiring the data comprises:for each slice of a plurality of slices, consecutively acquiring all shots of a multi-shot acquisition for the slice before acquiring data for a next slice, wherein each shot uses a radio frequency (RF) excitation pulse with a different flip angle; andreconstructing images from the acquired magnetic resonance data.

2. The method of claim 1 , wherein the images are reconstructed from the magnetic resonance data using a subspace reconstruction algorithm that utilizes basis vectors to represent signal evolution across an echo train.

3. The method of claim 2, wherein the subspace reconstruction algorithm utilizes three basis vectors extracted through principal component analysis.

4. The method of claim 3, wherein the basis vectors are derived by simulating signal evolution based on Bloch equation and acquisition parameters for the multi-shot acquisition.

5. The method of claim 1, wherein acquiring the data further comprises applying a spoiling gradient along a slice-select axis after each image echo readout to suppress unwanted echo refocusing.

6. The method of claim 1. wherein acquiring the data further comprises acquiring navigator echo data by sampling a navigator echo after each RF excitation pulse.20QBM25141.04924\99501997.1MGH 2024-148-02125141.049247. The method of claim 6. wherein the navigator echo comprises a three-line onedimensional navigator echo.

8. The method of claim 6, further comprising applying ghost correction to the acquired magnetic resonance data using the navigator echo data.

9. The method of claim 8, wherein the ghost correction comprises applying at least one of:intrashot first-order phase correction to each shot independently using the navigator echo data;intershot phase correction to match phases across different shots of a same slice; or intershot magnitude normalization to minimize sum-of-square differences between navigator magnitudes across different shots.

10. The method of claim 1. wherein each different flip angle is recursively determined according to 0;-i = tan 'fsin(Oi)). where i represents a shot number index of the multi-shot acquisition.

11. The method of claim 1. wherein the multi-shot acquisition comprises a 3-shot acquisition.

12. The method of claim 1, wherein the multi-shot acquisition comprises a 5-shot acquisition.

13. The method of claim 1, further comprising acquiring the magnetic resonance data using simultaneous multislice acceleration.

14. The method of claim 13, wherein the simultaneous multislice acceleration comprises applying gradient blips along a slice axis during readout.

15. The method of claim 14, wherein the gradient blips introduce a field-of-view shift in a phase-encoding direction across different slices.21QB\125141.04924\99501997.1MGH 2024-148-02125141.0492416. The method of claim 1. wherein the RF excitation pulses are designed to achieve consistent slice profiles across different shots in the multi-shot acquisition.

17. The method of claim 16, wherein the RF excitation pulses are designed using a recursive Shinnar-LeRoux algorithm.

18. The method of claim 1, further comprising applying complex-domain denoising to the reconstructed images to selectively remove thermal noise components.

19. The method of claim 1. further comprising generating neuronal activation maps from the reconstructed images.

20. The method of claim 1, further comprising generating functional connectivity maps from the reconstructed images.22QBM25141.04924\99501997.1

Images