Water-fat separated echo planar time-resolved imaging

Abstract

A method for water-fat separated magnetic resonance imaging (MRI) is provided herein. The method includes acquiring multi-echo MRI data from a subject using an echo planar time-resolved imaging (EPTI) pulse sequence, wherein echo spacing creates in-phase and out-of-phase relationships between water and fat signals. The method includes separating acquired data into odd echoes corresponding to in-phase data and even echoes corresponding to out-of-phase data, independently reconstructing the separated echoes to generate in-phase and out-of-phase images using subspace reconstruction, and generating separate water-only and fat-only multi-echo image series by performing water-fat separation using the in-phase images and the out-of-phase images.

Description

MGH 2024-337-02125141.04929WATER-FAT SEPARATED ECHO PLANAR TIME-RESOLVED IMAGING CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U. S. Provisional Patent Application Serial No. 63 / 723,444, filed on November 21, 2024, and entitled " SYSTEM AND METHOD OF ECHO PLANAR TIME-RESOLVED IMAGING,” which is herein incorporated by reference in its entirety.STATEMENT OF FEDERALLY SPONSORED RESEARCH

[0002] This invention was made with government support under AG083056 and NS 129893 awarded by the National Institutes of Health. The government has certain rights in the invention.BACKGROUND

[0003] Echo planar imaging (EPI) has been widely adopted in magnetic resonance imaging due to its rapid acquisition capabilities. However, conventional EPI techniques suffer from T2* blurring and geometric distortions, which limit their effectiveness in high-resolution imaging applications. Additionally, EPI is prone to chemical shift artifacts when fat suppression is insufficient, particularly in body imaging where fat tissues are present.

[0004] Echo planar time-resolved imaging (EPTI) addresses some of the distortion and blurring issues associated with conventional EPI by exploiting spatial-temporal correlations in the encoding process. EPTI can produce distortion-free and blurring-free multi-contrast images with high efficiency, making it suitable for various brain MRI applications including multiparametric quantitative imaging. However, the challenge of fat suppression and separation remains a barrier to broader application of EPTI techniques, particularly in body imaging where effective water-fat separation is needed to obtain high-quality images free from fat-related artifacts.SUMMARY OF THE DISCLOSURE

[0005] According to an aspect of the present disclosure, a method for water-fat separated magnetic resonance imaging (MRI) is provided. The method includes acquiring multi-echo magnetic resonance imaging data from a subject with an MRI system using an echo planar time-resolved imaging (EPTI) pulse sequence, where an echo spacing parameter is selected to create in-phase and out-of-phase relationships between water and fat signals. The 1QB\125141.04929\99569738.3MGH 2024-337-02125141.04929method includes separating the acquired multi-echo data into odd echoes corresponding to in-phase data and even echoes corresponding to out-of-phase data. The method includes independently reconstructing the separated odd echoes to generate in-phase images and the separated even echoes to generate out-of-phase images using subspace reconstruction.. The method includes generating separate water-only and fat-only multi-echo image series by performing water-fat separation using the in-phase images and the out-of-phase images.

[0006] According to another aspect of the present disclosure, an MRI system is provided. The MRI system includes a magnet assembly configured to generate a polarizing magnetic field, a gradient system configured to generate magnetic field gradients, a radiofrequency system configured to transmit radiofrequency pulses and receive magnetic resonance signals, and a computer system. The computer system is configured to control the gradient system and the radiofrequency system to perform an EPTI pulse sequence including a multi-echo readout train; an echo spacing set to a multiple of a period of a fat chemical shift frequency such that odd echoes acquire in-phase k-space data where water and fat signals are additive and even echoes acquire out-of-phase k-space data where water and fat signals are subtractive; separate spatiotemporal encoding trajectories for the odd echoes and the even echoes, where consecutively acquired pairs of odd and even echoes sample at same k-space positions; and a bipolar readout gradient to acquire the multi-echo data.BRIEF DESCRIPTION OF THE DRAWINGS

[0007] FIGS. 1A-1B illustrate examples of conventional echo planar time-resolved imaging (EPTI) pulse sequences and encoding patterns.

[0008] FIGS. 1C-1D illustrate examples of water-fat-separated (WFS) EPTI pulse sequences and encoding patterns with water-fat separation capabilities, according to aspects of the present disclosure.

[0009] FIG. IE illustrates an example auto-calibrated encoding pattern of hybrid k-t space with water-fat separation capabilities, according to aspects of the present disclosure.

[0010] FIG. 2 illustrates a flowchart of an example method for WFS-EPTI data acquisition and image reconstruction, according to aspects of the present disclosure.

[0011] FIG. 3 illustrates an example of separating multi-echo data into odd and even echo subsets.

[0012] FIG. 4 is a block diagram of an example MRI system configured to implement WFS-EPTI methods, according to aspects of the present disclosure.2QB\125141.04929\99569738.3MGH 2024-337-02125141.04929DETAILED DESCRIPTION

[0013] Described here are systems and methods for water-fat separated echo planar time-resolved imaging (EPTI) that combine the rapid acquisition and distortion-free capabilities of EPTI with effective water-fat separation techniques. The disclosed approaches may enable high-quality body imaging by addressing the longstanding challenge of fat-related artifacts in echo planar imaging (EPI) while maintaining the temporal efficiency and multicontrast capabilities of EPTI.

[0014] In some aspects, the disclosed systems and methods incorporate water-fat separation algorithms directly into the EPTI reconstruction process, allowing for simultaneous acquisition of water-only and fat-only images with reduced chemical shift artifacts. The systems and methods may utilize multi-echo acquisition strategies and iterative reconstruction techniques that exploit both spatial-temporal correlations and the known chemical shift properties of water and fat to achieve robust separation. In some cases, the disclosed techniques may provide improved image quality for body MR1 applications while preserving the rapid acquisition speeds that make EPTI attractive for time-sensitive imaging scenarios.

[0015] The water-fat separated echo planar time-resolved imaging (WFS-EPTI) techniques described in the present disclosure represent an advancement in MRI that addresses the challenge of fat suppression and separation in EPI sequences. Conventional EPI suffers from T2* blurring and geometric distortions, which can hinder high-resolution imaging applications. In some cases, conventional EPI may be prone to chemical shift artifacts when fat suppression is insufficient, particularly in body imaging applications where fat tissues are prevalent.

[0016] EPTI addresses the distortion and blurring limitations of conventional EPI by exploiting spatial-temporal correlations in the encoding process. In this way, EPTI may produce distortion-free and blurring-free multi-contrast images with high efficiency, making the technique suitable for various brain MRI applications including multi-parametric quantitative imaging. However, the challenge of fat suppression and separation may need to be addressed for broader application of EPTI in body imaging scenarios.

[0017] The WFS-EPTI techniques described in the present disclosure further enable effective fat-water separation while providing efficient distortion-free multi-contrast imaging capabilities. In some cases, the multi-echo data acquired by an EPTI readout may intrinsically encode the phase information of water and fat signals, making the technique suitable for waterfat separation applications. The WFS-EPTI approach may address several technical challenges,3QB\125141.04929\99569738.3MGH 2024-337-02125141.04929including the rapid phase evolution of fat signals across the EPTI readout that can compromise temporal correlation and violate low-rank assumptions in subspace reconstruction, potentially leading to degraded reconstruction performance.

[0018] In some cases, the bipolar echo-planar readout in EPTI may result in chemicalshift induced misregistration that may require correction during the water-fat separation process. The WFS-EPTI technique may incorporate an in-phase and out-of-phase acquisition and encoding scheme to ensure accurate subspace reconstruction while preserving acquisition efficiency. Additionally, the approach may adopt k-space-based water-fat separation methods to address issues caused by the bipolar readout configuration.

[0019] The WFS-EPTI system may provide an efficient method for fat separation and suppression, obtaining distortion-free water-only and fat-only multi-echo images along with quantitative parameters such as MRI relaxometry measurements including T2, T2*, and proton density values in reduced timeframes. The technique may facilitate distortion-free imaging in body MRI applications with capabilities for separating, removing, or quantifying fat signals. In some cases, WFS-EPTI may be applied across various imaging contrasts and applications, including multi-contrast imaging, quantitative imaging, and diffusion imaging sequences.

[0020] EPTI acquires data by sampling a hybrid space spanned by the phase encoding dimension and the temporal dimension using a zigzag trajectory. In some cases, the EPTI acquisition process utilizes a series of gradient echoes generated during a single excitation, where each echo in the train is encoded with different spatial-temporal information. Interleaved phase encoding lines can also be utilized such that temporally adjacent phase encoding lines are separated in time by a first temporal spacing and phase encoding lines that are adjacent in k-space are separated in time by a second temporal spacing that is greater than the first temporal spacing. In some cases, the multi-echo nature of EPTI provides temporal sampling of the MR signal decay, capturing both magnitude and phase evolution across the echo train while addressing geometric distortions and blurring common to conventional echo planar imaging.

[0021] The zigzag traj ectory in EPTI intrinsically encodes phase information from both water and fat signals due to the chemical shift difference between these tissue types. The zigzag pattern ensures that neighboring k-space points are acquired only a few milliseconds apart, containing small Bo-inhomogeneity induced phase and T2* decay that can be estimated well by parallel imaging and Bo-inhomogeneity-informed reconstruction. Water and fat protons resonate at slightly different frequencies, with fat typically resonating approximately 3.5 parts per million lower than water at clinical field strengths. This frequency difference results in a4QB\125141.04929\99569738.3MGH 2024-337-02125141.04929phase evolution between water and fat signals that accumulates over time during the echo train acquisition.

[0022] During the EPTI readout, the phase relationship between water and fat signals may evolve according to the chemical shift difference and the echo timing parameters. The gradient-echo image at each echo position contains contributions from both water and fat components, where the fat component exhibits a time-dependent phase term relative to the water component. This phase evolution can be described mathematically, where the image intensity at a given echo is expressed as a combination of water intensity, fat intensity' with chemical shift encoding, T2* signal decay, and BO-inhomogeneity-induced phase terms.

[0023] The chemical shift encoding term in the multi-echo data may depend on the fat frequency shift, the echo time of the first echo, and the echo spacing parameters. In some cases, the echo spacing may be defined as the time interval between consecutive echoes in the EPTI readout train. The T2* signal decay component accounts for the transverse relaxation effects that occur during the echo train, while the BO-inhomogeneity-induced phase captures magnetic field variations across the imaging volume. The rapid acquisition of multiple echoes in EPTI provides sufficient phase encoding information to enable water-fat separation algorithms while achieving distortion-free and blurring-free imaging. In some cases, the multi-echo data contains the temporal sampling needed to resolve the phase differences between water and fat signals, allowing for subsequent separation of these components during image reconstruction using tilted reconstruction kernels. The echo tram length and echo spacing parameters can be selected to provide adequate phase sampling while maintaining the rapid acquisition characteristics of the EPTI sequence.

[0024] The spatial-temporal encoding in EPTI is designed to exploit correlations in the multi-echo data while preserving the phase information needed for water-fat separation. In some cases, the encoding strategy balances the requirements for rapid acquisition, spatial resolution, and phase sampling to enable effective separation of water and fat components. The multi-echo acquisition provides the raw data foundation upon which water-fat separation algorithms operate to generate separate water-only and fat-only image series.

[0025] The WFS-EPTI method may be implement various pulse sequence configurations to address different imaging requirements and contrast mechanisms. As one example, gradient-echo sequences may be used for WFS-EPTI, where the multi-echo readout train may be generated through repeated application of frequency encoding gradients with alternating polarity. The gradient-echo configuration provides rapid acquisition capabilities5QB\125141.04929\99569738.3MGH 2024-337-02125141.04929while maintaining the phase information needed for water-fat separation, making this sequence type suitable for applications requiring high temporal resolution or large anatomical coverage.

[0026] Spin-echo sequences may incorporate WFS-EPTI techniques through implementation of refocusing pulses that may reverse the effects of static magnetic field inhomogeneities while preserving the chemical shift encoding needed for water-fat separation. The spin-echo implementation may provide T2-weighted contrast that may be less susceptible to magnetic susceptibility artifacts compared to gradient-echo approaches. The refocusing pulses in spin-echo sequences may enable imaging in regions with significant magnetic field variations while maintaining the multi-echo acquisition pattern required for effective water-fat separation.

[0027] Combined gradient-echo spin-echo sequences may utilize WFS-EPTI methods to simultaneously acquire both T2-weighted and T2*-weighted contrast information within a single acquisition. The combined sequence approach may generate multiple echo trains with different contrast characteristics, where early echoes may provide gradient-echo contrast and later echoes may provide spin-echo contrast following refocusing pulse application. The dual contrast acquisition may enable comprehensive tissue characterization while maintaining the water-fat separation capabilities throughout both portions of the echo train.

[0028] Stimulated echo sequences may employ WFS-EPTI techniques where the multiecho readout may be applied following the stimulated echo formation process. The stimulated echo configuration may utilize three radiofrequency pulses to generate echoes with specific timing characteristics that may be advantageous for certain tissue types or imaging applications. The water-fat separation methods may be applied to the stimulated echo signals to provide clean tissue separation while exploiting the unique contrast mechanisms available through stimulated echo formation.

[0029] Diffusion imaging applications may incorporate WFS-EPTI methods to address the challenge of fat contamination in diffusion-weighted acquisitions. Diffusion-weighted sequences may apply motion-sensitizing gradients that may encode molecular motion information into the signal phase and magnitude, while the multi-echo readout may provide the temporal sampling needed for water-fat separation. The combination of diffusion encoding and water-fat separation may enable clean diffusion parameter estimation from water tissues while eliminating artifacts from fat signal contributions. The diffusion-weighted WFS-EPTI implementation may apply diffusion gradients prior to the multi-echo readout train, where the diffusion encoding may affect both water and fat components according to their respective6QB\125141.04929\99569738.3MGH 2024-337-02125141.04929diffusion characteristics. The water-fat separation process may be applied to the diffusion-weighted multi-echo data to generate separate diffusion-weighted images for water and fat components. The separated water-only diffusion data may enable calculation of diffusion parameters such as apparent diffusion coefficient values without contamination from fat tissues that may exhibit different diffusion properties. The diffusion WFS-EPTI approach may be particularly valuable for body diffusion imaging applications such as liver, kidney, or prostate imaging where fat infiltration or adjacent fat tissues may compromise conventional diffusion measurements. The method may enable accurate diffusion quantification in these challenging anatomical regions while maintaining the rapid acquisition characteristics needed for breathhold or motion-robust imaging protocols.

[0030] Quantitative imaging sequences may utilize WFS-EPTI approaches to enable simultaneous measurement of multiple tissue parameters while addressing fat-related artifacts. Quantitative sequences may be designed to sample specific aspects of tissue relaxation, perfusion, or metabolic characteristics through controlled variation of sequence timing parameters or contrast preparation modules. The multi-echo nature of WFS-EPTI may provide the temporal sampling needed for quantitative parameter estimation while the water-fat separation may ensure that quantitative measurements reflect the intended tissue components.

[0031] WFS-EPTI can be used for various clinical applications, including brain imaging, body imaging, and the like. Brain imaging applications may benefit from WFS-EPTI implementation in regions where fat tissues may be present, such as in the scalp, orbital regions, or skull base areas. The method may enable distortion-free brain imaging while eliminating artifacts from subcutaneous fat or fat-containing anatomical structures that may otherwise contaminate the brain tissue signals. The rapid acquisition capabilities of WFS-EPTI may be particularly advantageous for brain imaging applications requiring high temporal resolution, such as functional MRI or dynamic contrast-enhanced studies.

[0032] Body imaging applications may benefit from WFS-EPTI implementation due to the prevalence of fat tissues throughout the body and the associated challenges with conventional fat suppression techniques. The method may enable high-quality body imaging in regions such as the abdomen, pelvis, thorax, and extremities where fat and water tissues may be intermixed or where conventional fat suppression may be ineffective. For abdominal imaging applications, strategies for addressing respiratory motion may include breath-hold acquisition protocols, respiratory triggering techniques, or retrospective motion correction methods. The breath-hold approach may utilize rapid acquisition capabilities of WFS-EPTI to7QB\125141.04929\99569738.3MGH 2024-337-02125141.04929complete imaging within a single breath-hold duration, thereby minimizing motion artifacts. Alternatively, respiratory triggering may synchronize data acquisition with specific phases of the respiratory cycle to ensure consistent anatomical positioning across different acquisition segments. Inter-shot motion correction techniques may be employed to detect and correct for motion between different shots of the acquisition, enabling free-breathing protocols that may be more comfortable for patients. The distortion-free characteristics of EPTI combined with effective water-fat separation may provide improved image quality for body imaging compared to conventional echo-planar approaches. The body imaging implementations of WFS-EPTI may address specific challenges associated with large field-of-view acquisitions where magnetic field inhomogeneities may compromise conventional fat suppression techniques. The WFS-EPTI approach may provide robust water-fat separation across large imaging volumes without requiring the uniform magnetic field conditions that may be needed for effective fat saturation pulses. The method may enable consistent fat suppression performance across different body regions and patient populations.

[0033] In some cases, WFS-EPTI may implement two-dimensional EPTI acquisitions for single-slice or multi-slice imaging protocols where each slice may be acquired independently with its own multi-echo readout train. The 2D implementation may provide rapid slice acquisition with effective water-fat separation, enabling efficient coverage of multiple anatomical locations or dynamic imaging of single slices with high temporal resolution. The 2D approach may be suitable for applications requiring real-time or near-realtime imaging capabilities.

[0034] In some other cases, WFS-EPTI may implement three-dimensional EPTI acquisitions for volumetric imaging where the multi-echo readout may be applied to 3D k-space encoding patterns. The 3D implementation may provide isotropic spatial resolution and comprehensive anatomical coverage while maintaining the water-fat separation capabilities throughout the imaging volume. The 3D approach may enable efficient whole-organ or wholebody imaging protocols with consistent water-fat separation performance across all spatial dimensions. The 3D WFS-EPTI implementation may utilize advanced k-space sampling strategies such as spiral or radial trajectories that may be optimized for the specific requirements of volumetric water-fat separation. The 3D encoding may enable acceleration techniques such as parallel imaging or compressed sensing that may further reduce acquisition times while preserving the multi-echo sampling needed for effective tissue separation. For motion-sensitive applications such as abdominal imaging, external calibration approaches may8QB\125141.04929\99569738.3MGH 2024-337-02125141.04929be employed where calibration data acquired during one breath-hold or motion state may be used for reconstruction of data acquired during different breath-holds or motion states. This external calibration strategy may provide improved motion robustness by allowing shorter acquisition times per breath-hold while maintaining reconstruction quality through the use of previously acquired calibration information. The volumetric approach may provide comprehensive tissue characterization across large anatomical regions in clinically acceptable scan times.

[0035] Referring now to FIGS. 1A-1D, example pulse sequences and encodings used by conventional EPTI (FIGS. 1 A and IB) and the proposed WFS-EPTI (FIGS. 1C and ID) are shown. In the illustrated example, a combined gradient-echo and spin-echo pulse sequence is used for WFS-EPTI.

[0036] As an example, when considering the chemical-shift of fat, the gradient-echo image Imat the (m+l)-th echo can be expressed as:m I rvi r F I

[0037] In this mathematical representation, pwrepresents the intensity of the water component at each spatial location, providing the magnitude contribution from water protons in the imaging volume. The fat component may be characterized by pr, which represents the intensity of the fat component with an associated frequency shift fr measured in Hz. This frequency shift may arise from the chemical shift difference between water and fat protons, typically corresponding to approximately 3.5 parts per million at clinical magnetic field strengths.

[0038] The temporal parameters in the signal model include to, which represents the echo time of the first echo in the EPTI readout train, and Tesp, which represents the echo spacing parameter defining the time interval between consecutive echoes. In some cases, these timing parameters may be selected to optimize the phase sampling characteristics while maintaining rapid acquisition speeds.-ilrt t^+mT^p 1

[0039] The chemical-shift encoding term, e captures the phase evolution of the fat signal relative to the water signal as a function of echo position m in the readout train. This term may exhibit rapid phase changes across the echo train due to the chemical shift difference between water and fat, with the phase accumulation increasing linearly with echo time. The magnitude of this phase evolution may depend on the fat frequency shift fr and the9QB\125141.04929\99569738.3MGH 2024-337-02125141.04929specific echo timing parameters used in the acquisition. The 72* signal decay is represented by -{tf+nilgsp -2e, which accounts for the transverse relaxation effects that occur during the echo train readout. This exponential decay term affects both water and fat components, with the decay rate determined by the T2* relaxation time constant of the respective tissues. In some cases, the T2* decay may vary spatially across the imaging volume due to tissue heterogeneity -i')6B(ta+mTesp\ and magnetic field variations. The BO-inhomogeneity-induced phase, e, captures phase variations arising from magnetic field inhomogeneities across the imaging volume. The parameter represents the gyromagnetic ratio, while AB represents the local magnetic field inhomogeneities. This phase term may contribute to spatial variations in signal phase that may need to be accounted for during the reconstruction process.-i2?rfF(tQ+mTesp^

[0040] The fast-evolving fat phase e may present challenges for subspace reconstruction algorithms commonly used in EPTI. In some cases, subspace reconstruction methods may rely on low-rank assumptions about the temporal behavior of the multi-echo data, where the signal evolution across echoes may be expected to follow smooth, predictable patterns that can be represented efficiently in a reduced-dimensional subspace. The rapid phase evolution of the fat signal may violate these low-rank assumptions by introducing high-frequency temporal variations that may not be well-represented in the A pical subspace basis functions. As the echo index m increases, the phase term may accumulate significant phase changes that can appear as high-frequency oscillations in the temporal domain. These oscillations may increase the rank of the data matrix, making the low-rank approximation less accurate and potentially degrading the reconstruction performance.

[0041] To resolve this problem, a Dixon method can be incorporated into the EPTI acquisition, where TEs and T are chosen such that the odd and even echoes acquire in-phase and out-of-phase images, respectively. Images acquired in WFS-EPTI can then be defined as:f (p„ +m E oddechoes4m s evenechoes

[0042] The in-phase and out-of-phase acquisition strategy in WFS-EPTI may therefore address the challenges posed by rapidly evolving fat phase terms through strategic selection of echo spacing parameters. For example, the echo spacing Tespmay be chosen as 1 / k times, or k times, the period of the fat chemical shift, where k is a positive integer. In some examples, k =10QB\125141.04929\99569738.3MGH 2024-337-02125141.049292, such as where the echo spacing is set to be one half the period of the fat chemical shift, corresponding to 1 / (2fr). where fp represents the fat frequency shift in Hz. This specific timing selection may result in a configuration where odd echoes acquire in-phase images and even echoes acquire out-of-phase images. The echo spacing may be selected based on the specific magnetic field strength to achieve the desired in-phase and out-of-phase relationship. For example, at 3 T, the echo spacing may be selected to approximately 1.1 milliseconds, while at 1.5 T the echo spacing may be approximately 2.2 milliseconds, and at 7 T the echo spacing may be approximately 0.47 milliseconds. Solutions for imperfect echo spacing and chemical shift frequency variations may include adaptive echo spacing adjustment based on measured field strength, incorporation of phase deviation terms in the subspace basis generation, or use of data-driven basis extraction that can accommodate deviations from ideal timing conditions. The reconstruction algorithm may be designed to handle small deviations from perfect in-phase and out-of-phase conditions through complex-valued basis functions that can capture residual phase variations. This timing selection may cause the fat phase evolution to follow a predictable pattern where the chemical shift phase term alternates between consistent values for odd and even echo positions. The strategic timing may effectively modulate the fat phase evolution to create two separate data sets with improved temporal correlation characteristics.

[0043] By eliminating the fast-evolving chemical-shift term, in-phase and out-of-phase images can be reconstructed separately with significantly improved reconstruction conditioning. The improved reconstruction conditioning may arise from the restoration of low-rank characteristics in both the odd and even echo data sets. Without the rapidly oscillating fat phase term, the temporal evolution of the signal may follow predictable patterns that can be efficiently represented in reduced-dimensional subspaces. This may enable more accurate subspace reconstruction with better preservation of image quality and reduced artifacts.

[0044] In some cases, the echo spacing may be chosen as multiples (e.g., k times, 1 / k times) of the fat chemical shift period, corresponding to integer multiples of 1 / fr, such that the phase of fat across all echoes remains consistent. This alternative timing selection may address fat artifacts in the resultant images without requiring a separate water-fat separation step, as the fat phase remains constant throughout the echo train. The consistent fat phase may prevent the temporal correlation degradation that can occur with rapidly evolving phase terms.

[0045] To improve the image quality', encoding patterns for WFS-EPTI can be further optimized by implementing a spatiotemporal CAIPI trajectory’ separately for odd and even echoes. As shown in FIG. ID, the spatiotemporal encoding patterns for WFS-EPTI may be11QB\125141.04929\99569738.3MGH 2024-337-02125141.04929optimized by implementing separate spatiotemporal trajectories for odd and even echoes. Using these separate trajectories allows for independent optimization of the k-space sampling patterns for the in-phase and out-of-phase data sets. In some cases, the odd echoes may follow one spatiotemporal sampling trajectory while the even echoes may follow a different trajectory, with both trajectories designed to provide complementary k-space coverage. These optimized encoding patterns help ensure that consecutively acquired pairs of odd and even echoes sample at the same k-space positions, providing matched spatial frequency information for the in-phase and out-of-phase data sets. This matched sampling may facilitate the subsequent water-fat separation process by ensuring that corresponding spatial frequency components are available from both echo types. The coordinated sampling strategy may maintain the spatial registration between in-phase and out-of-phase images while allowing for independent temporal optimization of each echo subset.

[0046] Variable density encoding strategies may also be implemented to achieve further acceleration by oversampling the central k-space region while undersampling the peripheral regions. The central k-space region may be densely sampled with fully sampled center lines and moderately accelerated adjacent regions, while the remaining periphery may be highly accelerated using the spatiotemporal trajectory. This variable density approach may enable higher acceleration factors while preserving image quality' by ensuring adequate sampling of low spatial frequency information that contributes significantly to image contrast and signal-to-noise ratio.

[0047] An example implementation of the WFS-EPTI techniques described in the present disclosure may be described with reference to FIG. 2, which illustrates an example workflow- from data acquisition through final image reconstruction. The process may begin at step 202 with the acquisition of undersampled EPTI data using echo spacing parameters selected as described above to improve water-fat separation. The multi-echo data acquisition may utilize the in-phase and out-of-phase timing scheme described above where the echo spacing Tespmay be set to k times or 1 / k times the period of the fat chemical shift, enabling the separation of water and fat signal components through controlled phase relationships.

[0048] At step 204, the acquired multi-echo data may undergo separation into distinct odd and even echo subsets based on the echo index positions within each readout train, as illustrated in FIG. 3. The separation process may partition the multi-echo data based on the echo index, where echoes at odd positions (1st, 3rd, 5th. etc.) may be grouped together to form the in-phase data set, while echoes at even positions (2nd, 4th, 6th, etc.) may be grouped to12QB\125141.04929\99569738.3MGH 2024-337-02125141.04929form the out-of-phase data set. This separation process may partition the temporal data into two independent datasets that may exhibit improved temporal correlation characteristics compared to the original combined dataset.

[0049] The separated odd and even echo subsets are then independently reconstructed, as indicated at steps 206 and 208, respectively. The separated odd echo data may proceed to step 206 for independent subspace reconstruction to generate in-phase images. The subspace reconstruction algorithm may exploit the restored temporal correlations in the odd echo subset to produce high-quality images representing the combined water and fat signal intensities. The reconstruction process may utilize basis functions optimized for the smooth temporal evolution patterns characteristic of the in-phase data. Simultaneously, or sequentially, the separated even echo data may proceed to step 208 for independent subspace reconstruction to generate out-of-phase images. The even echo reconstruction may operate on data containing the difference between water and fat signal intensities, utilizing subspace methods tailored to the specific signal characteristics of the out-of-phase acquisitions. The parallel processing of odd and even echo subsets may enable simultaneous reconstruction while maintaining computational efficiency.

[0050] The subspace reconstruction process may utilize a mathematical framework that exploits the temporal correlations present in the multi-echo data to recover high-quality images from undersampled acquisitions. The reconstruction may employ complex-valued subspace basis functions that can capture both magnitude and phase variations in the signal evolution, enabling robust handling of imperfect in-phase and out-of-phase conditions that may arise from non-ideal echo spacing or variations in fat chemical shift frequencies. In some aspects, the subspace reconstruction may begin by generating a set of basis functions that represent the expected signal evolution patterns across the echo train. These basis functions may be derived through principal component analysis of simulated signal evolution curves that account for the specific acquisition parameters, including T2* decay characteristics, Bo field variations, and the temporal sampling scheme. Alternatively, data-driven basis extraction may be employed where basis functions are derived directly from calibration data, allowing the reconstruction to adapt to the specific signal characteristics present in the acquired dataset, including deviations from ideal in-phase and out-of-phase conditions. The complex-valued. The basis functions may form a low-dimensional subspace that can efficiently represent the actual signal evolution, thereby reducing the degrees of freedom in the reconstruction problem from the total number13QB\125141.04929\99569738.3MGH 2024-337-02125141.04929of echo time points to the much smaller number of basis functions. For example, the subspace reconstruction may be computed by:mm\\UFSB(l)c — y\\% + 4J?(c)c

[0051] 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-t space dataset. The regularization term R(c) can be incorporated to further improve the conditioning, and A is the control parameter of the regularization. A locally low-rank regularization in image domain or Hankle matrix based regularization in k-space domain can be used. After solving for c, images across echo train can be produced by the product of < >c.

[0052] The reconstruction algorithm may then estimate coefficient maps that describe how each basis function contributes to the signal at each spatial location in the imaging volume. This estimation process may involve solving an optimization problem that minimizes the difference between the acquired undersampled data and a forward model prediction, while incorporating regularization terms to improve conditioning and signal-to-noise ratio. The forward model may account for coil sensitivity variations, Bo-induced phase evolution, Fourier encoding, and / or the undersampling pattern used during data acquisition. In some cases, locally low -rank constraints may be applied to further improve reconstruction performance. Once the coefficient maps are determined, the final images may be generated by computing the linear combination of the basis functions weighted by their respective coefficients, producing a complete set of multi-echo images with enhanced temporal resolution and reduced artifacts compared to conventional reconstruction approaches.

[0053] In some cases, the subspace reconstruction may implement a jointly regularized reconstruction of the in-phase and out-of-phase images. In these instances, the in-phase and out-of-phase data are solved separately to maintain their respective data consistency, but regularized jointly to leverage shared structural / contrast information between them for improved reconstruction conditioning and SNR, as detailed below:

[0054] where < >oddand < >evenare subspace basis generated for the odd and even echoes, respectively; coddand cevenare the corresponding coefficient maps; yoddand yevenare the acquired k-t space datasets for odd and even echoes. While enforcing the data consistency separately for the odd and even echoes to account for the presence of fat signals, the regularization term R(codd, ceven) is applied jointly for both sets of images.14QB\125141.04929\99569738.3MGH 2024-337-02125141.04929

[0055] The independent reconstruction of odd and even echo subsets enables each data set to be processed using reconstruction algorithms optimized for the specific signal characteristics of in-phase and out-of-phase acquisitions. The in-phase data set may exhibit temporal evolution patterns corresponding to the combined water and fat signal intensities, while the out-of-phase data set may display evolution patterns corresponding to the difference between water and fat intensities. The separate processing may allow for tailored reconstruction approaches that can exploit the distinct temporal correlation structures present in each subset.

[0056] As another example, GRAPPA-based reconstruction approaches may be applied to reconstruct the multi-echo EPTI images including both magnitude and phase signals from the separated echo data. The GRAPPA (GeneRalized Autocalibrating Partially Parallel Acquisitions) reconstruction method utilizes parallel imaging techniques to reconstruct the missing k-space information using coil sensitivity information from multiple receiver channels. In the context of WFS-EPTI, GRAPPA-based reconstruction may be applied independently to the odd and even echo subsets to generate separate in-phase and out-of-phase image series. The GRAPPA reconstruction process may involve calibration of coil sensitivity maps from fully sampled central k-space regions, followed by synthesis of missing k-space data points using linear combinations of acquired data from multiple coils. The reconstruction may preserve both magnitude and phase information across the multi-echo series, maintaining the phase relationships that may be utilized in subsequent water-fat separation algorithms. The parallel imaging acceleration may enable reduced acquisition times while preserving the temporal sampling needed for effective water-fat separation.

[0057] In some cases, such as illustrated in FIG. IE, the calibration data may be acquired as part of an auto-calibrated scan. In the illustrated example, the auto-calibrated WFS-EPTI embeds calibration within the imaging scan by employing a non-uniform undersampling pattern that oversamples the k-space center. Specifically, the central region of krt space is densely sampled, with fully sampled center lines and moderately accelerated adjacent regions, while the remaining periphery is highly accelerated using the spatiotemporal EPTI trajectory. During reconstruction, the central auto-calibration region is first recovered using GRAPPA at each echo time. These auto-calibration data, acquired under the same contrast and motion state as the imaging data, are then used for coil and Bo calibration and for data-driven subspace basis extraction.

[0058] The data-driven bases are bases extracted from the calibration data rather than a simulated dictionary, which are also designed to be complex-valued. They can capture signal15QB\125141.04929\99569738.3MGH 2024-337-02125141.04929fluctuations resulting from imperfect in-phase / out-of-phase conditions (e.g., non-ideal echospacing and chemical-shift frequencies) specific to the dataset. Therefore, using these bases can improve the performance of subspace reconstruction.

[0059] Alternatively, the model-based reconstruction approaches may be utilized to directly estimate water and fat parameters based on the signal model, reducing the number of unknowns in the reconstruction process to improve reconstruction accuracy. The model-based approach may incorporate the mathematical formulation of the water-fat signal model directly into the reconstruction algorithm, allowing for simultaneous estimation of water intensity pw, fat intensity7pF, T2* relaxation parameters, and Bo field map information from the multi-echo data. The model-based reconstruction may formulate the reconstruction problem as a nonlinear optimization where the unknown parameters may be estimated by minimizing the difference between the acquired data and the predicted signal based on the water-fat model. The optimization process may incorporate regularization terms that promote spatial smoothness in the parameter maps while maintaining fidelity to the acquired data. The direct parameter estimation approach may reduce the dimensionality of the reconstruction problem compared to methods that first reconstruct individual echo images and subsequently perform water-fat separation. The reduced number of unknowns in model-based reconstruction may improve the conditioning of the inverse problem, particularly in cases where the multi-echo data may be undersampled or corrupted by noise. By constraining the reconstruction to follow the known signal model, the approach may provide more robust parameter estimation compared to sequential reconstruction and separation approaches. The model-based framework may also enable incorporation of prior knowledge about tissue properties, such as expected ranges for T2* values or spatial smoothness constraints on parameter maps.

[0060] As yet another example, machine learning-based approaches may be applied to extract water and fat images from multi-echo EPTI data for water-fat separation applications. The machine learning methods may utilize trained neural networks or other learning algorithms to directly map from multi-echo input data to separated water and fat output images. The learning-based approach may be trained on datasets containing multi-echo EPTI acquisitions with corresponding ground truth water and fat images obtained through reference methods. The machine learning reconstruction may employ deep neural network architectures designed to process multi-echo time series data and generate separated tissue component images. The network architecture may include convolutional layers for spatial feature extraction, recurrent layers for temporal processing of the echo train data, or attention mechanisms for adaptive16QB\125141.04929\99569738.3MGH 2024-337-02125141.04929weighting of different echo contributions. The training process may optimize network parameters to minimize reconstruction errors between predicted and reference water-fat separations across diverse imaging scenarios. The machine learning approach may offer computational advantages by providing rapid water-fat separation once the network has been trained, potentially enabling real-time or near-real-time processing of WFS-EPTI data. The learned mapping may capture complex relationships between multi-echo signal patterns and tissue composition that may be difficult to model analytically. In some cases, the machine learning method may be robust to variations in acquisition parameters, noise levels, or anatomical differences that might challenge traditional model-based approaches.

[0061] At step 210, the reconstructed in-phase and out-of-phase images may be transformed from the image domain to the k-space domain through two-dimensional Fourier transform operations. This transformation may prepare the data for k-space-based water-fat separation algorithms while preserving the magnitude and phase information from the reconstruction process.

[0062] The process may continue to step 212 where the k-space data from both in-phase and out-of-phase reconstructions may be combined to create a unified dataset for water-fat separation processing. The data combination may ensure that corresponding spatial frequency components from both echo subsets may be properly aligned and available for the separation algorithms. The combined k-space data may maintain the phase relationships needed for effective tissue component separation.

[0063] At step 214, k-space-based water-fat separation may be performed on the combined data to address chemical-shift induced misregistration effects caused by the bipolar readout configuration. Alternatively, the k-space-based separation method may be applied to both the in-phase and out-of-phase echo subsets, generating separate water and fat image series for each subset. The separated results from both subsets may be combined or processed independently depending on the specific application requirements. In some cases, the availability of both in-phase and out-of-phase separated data may provide additional information for quality assessment or parameter estimation accuracy evaluation.

[0064] The separation algorithm may apply echo-specific phase corrections that may compensate for the alternating spatial displacement pattern between water and fat components. For example, the bipolar echo-planar readout in EPTI may result in chemical-shift induced misregistration between water and fat components due to the alternating gradient polarity during the readout process. During bipolar readout, the frequency encoding gradient may17QB\125141.04929\99569738.3MGH 2024-337-02125141.04929alternate direction between positive and negative polarities for consecutive echo acquisitions. This alternating gradient polarity may cause water and fat signals to be spatially shifted in opposite directions along the frequency encoding axis due to their different resonance frequencies. The chemical shift difference between water and fat may manifest as a spatial displacement in the frequency encoding direction, with the magnitude of displacement depending on the chemical shift value, the readout bandwidth, and the specific gradient timing parameters. In bipolar readout sequences, this displacement may alternate direction between consecutive echoes, resulting in a spatial misregistration pattern that varies across the echo train. The misregistration may cause water and fat components to appear at slightly different spatial locations in alternating echoes, complicating the separation process. Chemical shift misregistration correction strategies may be implemented by applying echo-specific spatial shift corrections in k-space that account for the acquisition timing differences between odd and even echo readout samples. The correction may involve applying different phase ramps to odd and even echoes based on the known chemical shift displacement and the specific bipolar gradient pattern, thereby restoring proper spatial alignment between water and fat components across all echoes in the acquisition.

[0065] The k-space-based water-fat separation method may address the chemical-shift induced misregistration by applying appropriate phase corrections and spatial shift compensations in the frequency domain. The k-space approach may utilize the linear phase relationship between spatial shifts and k-space phase to correct for the alternating displacement pattern. In some cases, the correction may involve applying echo-specific phase ramps in k-space that compensate for the spatial shifts caused by the bipolar readout pattern. The phase correction process may account for the alternating nature of the chemical shift displacement by applying different correction terms to odd and even echoes based on the gradient polarity used during acquisition. The correction terms may be calculated based on the known chemical shift value, the readout parameters, and the specific bipolar gradient pattern employed in the EPTI sequence. The k-space corrections may restore proper spatial registration between water and fat components across all echoes in the acquisition.

[0066] In some cases, the water-fat separation technique may incorporate multi-peak fat spectral models that account for the multiple resonance frequencies present in fat tissues, rather than assuming a single fat peak. The multi-peak modeling may consider the relative amplitudes and frequency offsets of different fat spectral components, enabling more accurate representation of the complex fat signal evolution across the echo train.18QB\125141.04929\99569738.3MGH 2024-337-02125141.04929

[0067] Advantageously, the k-space-based approach can provide improved separation accuracy compared to image-domain methods by addressing the spatial misregistration effects before the separation process. The frequency domain corrections may ensure that water and fat components are properly aligned across all echoes, enabling more accurate fitting of the signal evolution patterns. The corrected alignment may reduce artifacts and improve the quality of the separated water-only and fat-only images.

[0068] As indicated at step 216, separate water-only and fat-only multi-echo image series are generated through inverse Fourier transformation of the separated k-space data back to the image domain. The resulting distortion-free images may maintain the spatial resolution and contrast characteristics of the original EPTI acquisition while providing clean separation of water and fat tissue components. The multi-echo nature of the separated data may enable subsequent quantitative parameter mapping, including T2, T2*, proton density, and Bo field map calculations from both the water-only and fat-only image series.

[0069] By way of example, the multi-echo nature of the water-only data may provide temporal sampling of the transverse relaxation decay, allowing for quantitative estimation of T2* relaxation times at each spatial location in the imaging volume. The T2* parameter mapping may be accomplished through exponential fitting of the water signal magnitude as a function of echo time, where the decay rate may correspond to the inverse of the T2* relaxation time constant. The T2* fitting process may utilize nonlinear least squares optimization to estimate the T2* values from the multi-echo water-only magnitude data. The fitting algorithm may model the signal decay as an exponential function of echo time, accounting for the T2* relaxation while excluding contributions from fat components that have been separated during the reconstruction process. The water-only data may provide cleaner T2* estimates compared to conventional multi-echo approaches that may be contaminated by fat signal contributions or chemical shift artifacts.

[0070] T2 relaxation parameter maps may also be derived when the WFS-EPTI acquisition includes both gradient-echo and spin-echo components in a combined gradientecho spin-echo sequence configuration. The spin-echo portion of the acquisition may provide T2-weighted contrast that reflects the transverse relaxation time constant T2, which may be longer than T2* due to the refocusing of static magnetic field inhomogeneities. The T2 parameter estimation may involve fitting the signal evolution from the spin-echo portion of the multi-echo train, utilizing the water-only images to avoid contamination from fat signal contributions. The T2 fitting process may employ similar exponential decay modeling19QB\125141.04929\99569738.3MGH 2024-337-02125141.04929approaches as used for T2* estimation, but applied to the spin-echo derived signals that may exhibit slower decay rates. The availability of both T2 and T2* measurements from the same acquisition may enable calculation of additional parameters such as the apparent transverse relaxation rate, which may reflect the contribution of microscopic magnetic field variations to the observed signal decay.

[0071] Proton density maps may be calculated from the multi-echo water-only images by extrapolating the fitted exponential decay curves back to zero echo time to estimate the initial signal amplitude. The proton density parameter may represent the concentration of mobile water protons in each voxel, providing a measure of tissue water content that may be independent of relaxation effects. The extrapolation process may account for the T2 or T2* decay characteristics determined from the multi-echo fitting to provide accurate proton density estimates. The proton density calculation may utilize the fitted parameters from the T2 or T2* analysis to determine the y-intercept of the exponential decay function, which may correspond to the signal amplitude at zero echo time. This amplitude may be proportional to the proton density after appropriate calibration and correction for acquisition parameters such as flip angle, repetition time, and receiver gain settings. The water-only nature of the separated data may provide proton density maps that specifically reflect water content without contamination from fat protons.

[0072] Bo magnetic field maps may be derived from the phase evolution of the multiecho water-only images across the echo train. The Bo field inhomogeneities may cause linear phase accumulation as a function of echo time, where the slope of the phase evolution may be proportional to the local magnetic field deviation from the nominal field strength. The Bo mapping process may involve unwrapping the phase data across echoes and fitting linear functions to determine the phase evolution rate at each spatial location. The Bo field map calculation may utilize the relationship between phase accumulation and magnetic field variations, where the phase change per unit time may be proportional to the gyromagnetic ratio multiplied by the field deviation. The linear fitting of phase versus echo time may provide estimates of the local Bo field variations across the imaging volume. The water-only phase data may provide cleaner Bo field estimates compared to conventional approaches that may be affected by chemical shift phase contributions from fat tissues.

[0073] The quantitative parameter maps may be calculated using parallel processing approaches that can simultaneously fit multiple parameters from the multi-echo data. The fitting algorithms may incorporate constraints or regularization terms that promote spatial20QB\125141.04929\99569738.3MGH 2024-337-02125141.04929smoothness in the parameter maps while maintaining fidelity to the measured signal evolution. The simultaneous estimation approach may improve the robustness of parameter determination by utilizing the correlations between different quantitative measures.

[0074] The accuracy of the quantitative parameter maps may benefit from the distortion-free nature of the EPTI acquisition and the effective water-fat separation provided by the WFS-EPTI approach. The elimination of geometric distortions may ensure proper spatial correspondence between parameter maps and anatomical structures, while the water-fat separation may prevent contamination of quantitative measurements by unwanted signal contributions from fat tissues. The combination of rapid acquisition, distortion correction, and effective tissue separation may enable high-quality quantitative imaging suitable for clinical and research applications.

[0075] Additionally or alternatively, fat signal quantification can be performed using the reconstructed images, beyond simple separation or removal of fat contributions. The method may enable quantitative assessment of fat content, fat distribution, or fat composition through analysis of the separated fat-only image series. The fat quantification may provide clinically relevant information for applications such as hepatic steatosis assessment, muscle fat infiltration evaluation, or body composition analysis. The quantitative fat assessment may utilize the multi-echo fat-only images to derive fat-specific parameters such as fat T2* values, fat proton density, or fat spectral characteristics. The fat parameter mapping may provide insights into fat tissue properties that may be relevant for disease assessment or treatment monitoring. The simultaneous availability of both water and fat quantitative information may enable comprehensive tissue characterization that accounts for both tissue components. Fat removal may enable generation of water-only images that may be free from fat-related artifacts while preserving the underlying tissue contrast and spatial resolution characteristics. The fat removal may be particularly advantageous for applications where fat signal contributions may obscure pathological findings or interfere with automated image analysis algorithms. The clean water-only images may provide improved visualization of anatomical structures and pathological changes.

[0076] Referring particularly now to FIG. 4, an example of an MRI system 400 that can implement the methods described here is illustrated. The MRI system 400 includes an operator workstation 402 that may include a display 404, one or more input devices 406 (e.g., a keyboard, a mouse), and a processor 408. The processor 408 may include a commercially available programmable machine running a commercially available operating system. The21QB\125141.04929\99569738.3MGH 2024-337-02125141.04929operator workstation 402 provides an operator interface that facilitates entering scan parameters into the MRI system 400. The operator workstation 402 may be coupled to different servers, including, for example, a pulse sequence server 410, a data acquisition server 412, a data processing server 414, and a data store server 416. The operator workstation 402 and the servers 410, 412, 414, and 416 may be connected via a communication system 440, which may include wired or wireless network connections.

[0077] The pulse sequence server 410 functions in response to instructions provided by the operator workstation 402 to operate a gradient system 418 and a radiofrequency (“RF”) system 420. Gradient waveforms for performing a prescribed scan are produced and applied to the gradient system 418. which then excites gradient coils in an assembly 422 to produce the magnetic field gradients Gx, Gy, and Gzthat are used for spatially encoding magnetic resonance signals. The gradient coil assembly 422 forms part of a magnet assembly 424 that includes a polarizing magnet 426 and a whole-body RF coil 428.

[0078] RF waveforms are applied by the RF system 420 to the RF coil 428, or a separate local coil to perform the prescribed magnetic resonance pulse sequence. Responsive magnetic resonance signals detected by the RF coil 428, or a separate local coil, are received by the RF system 420. The responsive magnetic resonance signals may be amplified, demodulated, filtered, and digitized under direction of commands produced by the pulse sequence server 410. The RF system 420 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 410 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 428 or to one or more local coils or coil arrays.

[0079] The RF system 420 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 428 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:A / =7 / 2+02;

[0080] and the phase of the received magnetic resonance signal may also be determined according to the following relationship:22QB\125141.04929\99569738.3MGH 2024-337-02125141.04929

[0081] The pulse sequence server 410 may receive patient data from a physiological acquisition controller 430. By way of example, the physiological acquisition controller 430 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 410 to synchronize, or “gate,” the performance of the scan with the subject’s heart beat or respiration.

[0082] The pulse sequence server 410 may also connect to a scan room interface circuit 432 that receives signals from various sensors associated with the condition of the patient and the magnet system. Through the scan room interface circuit 432, a patient positioning system 434 can receive commands to move the patient to desired positions during the scan.

[0083] The digitized magnetic resonance signal samples produced by the RF system 420 are received by the data acquisition server 412. The data acquisition server 412 operates in response to instructions downloaded from the operator workstation 402 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 412 passes the acquired magnetic resonance data to the data processor server 414. In scans that require information derived from acquired magnetic resonance data to control the further performance of the scan, the data acquisition server 412 may be programmed to produce such information and convey it to the pulse sequence server 410. For example, during pre-scans, magnetic resonance data may be acquired and used to calibrate the pulse sequence performed by the pulse sequence server 410. As another example, navigator signals may be acquired and used to adjust the operating parameters of the RF system 420 or the gradient system 418, or to control the view order in which k-space is sampled. In still another example, the data acquisition server 412 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 412 may acquire magnetic resonance data and processes it in real-time to produce information that is used to control the scan.

[0084] The data processing server 414 receives magnetic resonance data from the data acquisition server 412 and processes the magnetic resonance data in accordance with instructions provided by the operator workstation 402. Such processing may include, for 23QB\125141.04929\99569738.3MGH 2024-337-02125141.04929example, 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 filters to raw k-space data or to reconstructed images, generating functional magnetic resonance images, or calculating motion or flow images.

[0085] Images reconstructed by the data processing server 414 are conveyed back to the operator workstation 402 for storage. Real-time images may be stored in a data base memory cache, from which they may be output to operator display 402 or a display 436. Batch mode images or selected real time images may be stored in a host database on disc storage 438. When such images have been reconstructed and transferred to storage, the data processing server 414 may notify the data store server 416 on the operator workstation 402. The operator workstation 402 may be used by an operator to archive the images, produce films, or send the images via a network to other facilities.

[0086] The MRI system 400 may also include one or more networked workstations 442. For example, a networked workstation 442 may include a display 444, one or more input devices 446 (e.g., a keyboard, a mouse), and a processor 448. The networked workstation 442 may be located within the same facility as the operator workstation 402, or in a different facility, such as a different healthcare institution or clinic.

[0087] The networked workstation 442 may gain remote access to the data processing server 414 or data store server 416 via the communication system 440. Accordingly, multiple networked workstations 442 may have access to the data processing server 414 and the data store server 416. In this manner, magnetic resonance data, reconstructed images, or other data may be exchanged between the data processing server 414 or the data store server 416 and the networked workstations 442, such that the data or images may be remotely processed by a networked workstation 442.

[0088] 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.24QB\125141.04929\99569738.3

Claims

MGH 2024-337-02125141.04929CLAIMS1. A method for water-fat separated magnetic resonance imaging (MRI), comprising:acquiring multi-echo magnetic resonance imaging data from a subject with an MRI system using an echo planar time-resolved imaging (EPTI) pulse sequence, wherein an echo spacing parameter is selected to create in-phase and out-of-phase relationships between water and fat signals;separating the acquired multi-echo data into odd echoes corresponding to in-phase data and even echoes corresponding to out-of-phase data;independently reconstructing the separated odd echoes to generate in-phase images and the separated even echoes to generate out-of-phase images using subspace reconstruction; andgenerating separate water-only and fat-only multi-echo image series by performing water-fat separation using the in-phase images and the out-of-phase images.

2. The method of claim 1, wherein the echo spacing parameter is selected to be 1 / k times a period of a fat chemical shift to create a constant phase within each echo group of the multi-echo data, wherein k is a positive integer.

3. The method of claim 1, wherein the echo spacing parameter is selected to be k times a period of a fat chemical shift to create a constant phase within each echo group of the multi-echo data, wherein k is a positive integer.

4. The method of claim 1, wherein generating the in-phase images and the out-of-phase images comprises establishing a substantially constant phase relationship between echoes, including echoes that have offsets from in-phase or out-of-phase echo times.

5. The method of claim 1, wherein the independent reconstruction of the in-phase images and the out-of-phase images includes using a jointly regularized reconstruction.

6. The method of claim 1, wherein the EPTI pulse sequence comprises a gradient-echo sequence.25QB\125141.04929\99569738.3MGH 2024-337-02125141.049297. The method of claim 1. wherein the EPTI pulse sequence comprises a spinecho sequence.

8. The method of claim 1, wherein the EPTI pulse sequence comprises a combined gradient-echo spin-echo sequence.

9. The method of claim 1, wherein the water-fat separation is a k-space water-fat separation comprising:transforming the reconstructed in-phase images and out-of-phase images to a k-space domain; andgenerating separated k-space data by performing k-space-based water-fat separation on the transformed data to reduce chemical-shift induced misregistration caused by bipolar readout gradients used in the EPTI pulse sequence.

10. The method of claim 9, wherein the k-space-based water-fat separation applies echo-specific phase corrections to compensate for alternating spatial displacement patterns between water and fat components.

11. The method of claim 1, wherein the water- fat separation is performed in an image domain.

12. The method of claim 1, wherein the EPTI pulse sequence implements separate spatiotemporal trajectories for the odd echoes and the even echoes.

13. The method of claim 12, wherein consecutively acquired pairs of odd and even echoes sample at the same k-space positions.

14. The method of claim 1, further comprising generating quantitative parameter maps from the separated water-only and fat-only multi-echo image series.

15. The method of claim 14, wherein the quantitative parameter maps comprise at least one of T2 relaxation maps, T2* relaxation maps, proton density maps, or Bo field maps.26QB\125141.04929\99569738.3MGH 2024-337-02125141.0492916. A magnetic resonance imaging (MRI) system, comprising:a magnet assembly configured to generate a polarizing magnetic field;a gradient system configured to generate magnetic field gradients;a radiofrequency system configured to transmit radiofrequency pulses and receive magnetic resonance signals; anda computer system configured to control the gradient system and the radiofrequency system to perform an echo planar time-resolved imaging (EPTI) pulse sequence comprising:a multi -echo readout train;an echo spacing set to a multiple of a period of a fat chemical shift such that odd echoes acquire in-phase k-space data where water and fat signals are additive and even echoes acquire out-of-phase k-space data where water and fat signals are subtractive;separate spatiotemporal encoding trajectories for the odd echoes and the even echoes, wherein consecutively acquired pairs of odd and even echoes sample at same k-space positions; anda bipolar readout gradient to acquire multi-echo data from the multi-echo readout train.

17. The MRI system of claim 16, wherein the echo spacing is equal to approximately 1 / k times a period of a fat chemical shift, wherein k is a positive integer.

18. The MRI system of claim 16, wherein the echo spacing is equal to approximately k times a period of a fat chemical shift, wherein k is a positive integer.

19. The MRI system of claim 17 or 18, wherein the polarizing magnetic field has a field strength of 3 Tesla and the period of the fat chemical shift is approximately 2.2 milliseconds.

20. The MRI system of claim 17 or 18, wherein the polarizing magnetic field has a field strength of 1.5 Tesla and the period of the fat chemical shift is approximately 4.4 milliseconds.27QB\125141.04929\99569738.3MGH 2024-337-02125141.0492921. The MRI system of claim 19 or 20, wherein k is equal to 2.

22. The MRI system of claim 17 or 18, wherein the polarizing magnetic field has an arbitrary field strength.

23. The MRI system of claim 16, wherein the EPTI pulse sequence comprises a gradient-echo sequence.

24. The MRI system of claim 16, wherein the EPTI pulse sequence comprises a spin-echo sequence.

25. The MRI system of claim 16, wherein the EPTI pulse sequence comprises a combined gradient-echo spin-echo sequence.

26. The MRI system of claim 25, wherein the combined gradient-echo spin-echo sequence generates multiple echo trains with different contrast characteristics.

27. The MRI system of claim 16, wherein the separate spatiotemporal encoding trajectories comprise CAIP1 trajectories.

28. The MRI system of claim 16, wherein the bipolar echo-planar readout gradient alternates gradient polarity between consecutive echoes to cause chemical-shift induced spatial displacement that alternates direction between the odd echoes and the even echoes.28QB\125141.04929\99569738.3

Images