Method for simultaneous multiple contrast magnetic resonance imaging of solid and soft tissue

Abstract

Methods, systems, and computer readable media for magnetic resonance imaging are disclosed. An example method for magnetic resonance (MR) imaging includes during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject. The method further includes receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject. The method further includes receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject. The method further includes generating and transmitting into the MR imaging subject a magnetic gradient for encoding during reception of the first and second echo signals.

Description

[0001] METHOD FOR SIMULTANEOUS MULTIPLE CONTRAST MAGNETIC RESONANCE IMAGING OF SOLID AND SOFT TISSUE

[0002] PRIORITY CLAIM

[0003] This application claims the priority benefit of U. S. Provisional Patent Application Serial No. 63 / 747,212, filed January 20, 2025, the disclosure of which is incorporated herein by reference in its entirety.

[0004] STATEMENT OF GOVERNMENT INTEREST

[0005] This invention was made with government support under DE028417, AR076392, AR068382, and AR050068 awarded by the National Institutes of Health, and 2026906, and 1845298 awarded by the National Science Foundation. The government has certain rights in the invention.

[0006] TECHNICAL FIELD

[0007] The subject matter described herein relates to magnetic resonance imaging. More particularly, the subject matter described herein relates to simultaneous multiple contrast resonance imaging of solid and soft tissue.

[0008] BACKGROUND

[0009] Computed tomography (CT) imaging is rapid, high-resolution, and produces images that are highly specific to bone. However, it uses ionizing radiation and is known to increase the lifetime risk of cancer, especially for patients with longer life expectancies. Furthermore, CT images exhibit poor contrast when visualizing the less attenuating soft tissues of the body, making it relatively insensitive to detecting and assessing pathologies in these compartments. Magnetic resonance imaging (MRI) does not use ionizing radiation and can produce soft tissue images of multiple contrasts, but to acquire multiple contrasts, several scans (or sequences) need to be run in series (back-to-back) in a single MR imaging protocol, prolonging the total duration of clinical examinations.Accordingly, there exists a need for improved methods, systems and computer readable media for imaging, including for magnetic resonance imaging.

[0010] SUMMARY

[0011] Methods, systems, and computer readable media for magnetic resonance imaging are disclosed. An example method for magnetic resonance (MR) imaging includes during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject. The method further includes receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject. The method further includes receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject. The method further includes generating and transmitting into the MR imaging subject a magnetic gradient for encoding during reception of the first and second echo signals. The method further includes generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times. The method further includes repeating the above listed steps for a plurality of repetition times that form the MR imaging scan. The method further includes simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and / or solid tissues.

[0012] In some embodiments, the presently disclosed subject matter provides a system for controlling a magnetic resonance imaging (MRI) scanner. Anexample system for controlling a magnetic resonance imaging (MRI) scanner includes at least one processor and a memory; and an MR imaging controller implemented by the at least one processor for controlling the MRI scanner to perform the following steps: during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject; receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject; receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject; generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals; generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times; repeating the above-listed steps for a plurality of repetition times that form the MR imaging scan; and simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and / or solid tissues.

[0013] An example non-transitory computer readable medium having stored thereon executable instructions that when executed by a processor of a computer control the computer to cause an MRI scanner to perform steps comprising, during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject. The steps further include receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject. The steps further include receiving, during the repetition time, a second echosignal at a second time after transmission of the RF pulse into the MR imaging subject. The steps further include generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals. The steps further include generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times. The steps further include repeating the above-listed steps for a plurality of repetition times that form the MR imaging scan. The steps further include simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective image of the MR imaging subject of soft and / or solid tissues.

[0014] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, the RF pulses follow an RF phase schedule φ(n) determined by the RF phase increment θ, where n is an index corresponding to the repetition time and θ ranging from greater than 0 degrees to less than about 5 degrees, optionally about.5° to about 3°. According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, the RF phase schedule follows the formula φ(n)=φ(n−1)+nθ, where φ(n) is the phase of the RF pulse for the nth repetition time and φ(n − 1) is the phase of the RF pulse for the (n-1)th repetition time.

[0015] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, receiving the first echo signal includes receiving the first echo signal at a time within a range of about 10 μs to about 100 μs after transmission of the RF pulse and receivingthe second echo signal includes receiving the second echo signal at a time within a range of about 1 ms to about 4 ms after transmission of the RF pulse.

[0016] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images comprises applying phase modulation, optionally according to a phase modulation angle i according to the following equation:

[0017]

[0018] wherein

[0019]

[0020] denotes the extracted contrast, which was encoded along an angle i measured from the real axis. ℜ(u) denotes the voxel-wise real value of an arbitrary complex-valued image u, and |u| is the voxel-wise absolute value of u.

[0021] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, a background phase is estimated with two acquisitions, one with a positive RF phase increment and one with equal but negative RF phase increment. In some embodiments, images of the MR are corrected with the background phase.

[0022] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, the T1 -weighted images of the MR imaging subject pertain to soft tissue; T2-weighted images of the MR imaging subject pertain to soft tissue and / or bone tissue; and a short T2-selective image of the MR imaging subject pertains to bone tissue.

[0023] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, the MR imaging subject is a pediatric subject.

[0024] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, the MR images comprise accelerated, high-resolution, and / or self-registered MR images.

[0025] The subject matter described herein can be implemented in software in combination with hardware and / or firmware. For example, the subject matter described herein can be implemented in software executed by a processor. In one exemplary implementation, the subject matter describedherein can be implemented using a non-transitory computer readable medium having stored thereon computer executable instructions that when executed by the processor of a computer control the computer to perform steps. Exemplary computer readable media suitable for implementing the subject matter described herein include non-transitory computer-readable media, such as disk memory devices, chip memory devices, programmable logic devices, and application specific integrated circuits. In addition, a computer readable medium that implements the subject matter described herein may be located on a single device or computing platform or may be distributed across multiple devices or computing platforms.

[0026] Accordingly, it is an object of the presently disclosed subject matter to provide methods, systems, and computer readable media for simultaneous multiple contrast magnetic resonance imaging of solid and soft tissue.

[0027] An object of the presently disclosed subject matter having been stated hereinabove, and which is achieved in whole or in part by the presently disclosed subject matter, other objects will become evident as the description proceeds hereinbelow.

[0028] BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The subject matter described herein will now be explained with reference to the accompanying drawings of which:

[0030] FIGS. 1A and 1B: Processing steps of partially spoiled gradient echo (PSGRE) and dual-echo ultrashort echo (UTE) pulse sequences, the parent sequences of DREAMER. FIG. 1A: PSGRE sequences acquire multi-contrast images with tunable Ti- and T2-weighted contrast. Positive and negative passes are used to remove the background phase and recover the multicontrast image. The phase modulation parameter ψ dictates contrast extraction of T1w and T2w images. Modulation of ψ adjusts the soft tissue contrast. FIG 1B: Dual-echo UTE sequences acquire two images with and without bone signal (ultrashort and conventional echo images, respectively). A normalized echo subtraction selectively enhances bone signal and suppresses soft tissues.FIG. 2A: Pulse sequence diagram of DREAMER (Dual Repetition and Echo Acquisition with Multi-contrast Encoding and Reconstruction). DREAMER combines the image contrast mechanisms of the PSGRE and dual-echo UTE parent sequences. Each RF pulse in DREAMER is followed by two echo acquisitions (one ultrashort, one conventional) to sensitize the sequence to short-T2 (bone-selective) signal. The phase schedule φ(n) uses a small increment to encode T2-weighting along the signal phase for long-T2 (soft tissue) species. Two repetitions (one positive and one negative RF phase increment) are acquired to correct for the background phase. FIG. 2B: A single DREAMER scan produces four datasets (2 echoes x 2 passes), which are used to compute two multi-contrast images (ultrashort and conventional echo).

[0031] FIGS. 3A and 3B: Image reconstruction of solid and soft tissue image contrasts from a single MRI scan with DREAMER. FIG. 3A: DREAMER produces ultrashort and conventional echo multi-contrast images with and without bone-selective signal. The vertical phase scale gives the values for the image phase in radians. FIG. 3B: From the two multi-contrast images all image contrasts can be computed. Phase modulation ψ enables tunable soft tissue contrast extraction of T1w and T2w images. A normalized echo subtraction between the two multi-contrast images yields a bone-selective image in which short-T2 species are enhanced.

[0032] FIGS. 4A and 4B: All representative contrasts of DREAMER for the three principal anatomical planes (24 years old (y.o.) healthy female). FIG.

[0033] 4A: A single DREAMER acquisition produces two three-dimensional multicontrast images (ultrashort and conventional echoes). Although the multicontrast magnitude image is relatively featureless, Tiw and T2W images (ψT1= 117°, ψT2= 156°), encoded along the signal phase, can be derived from either multi-contrast image to substantially enhance the soft-tissue contrast. Moreover, a bone-selective image can be derived by performing a normalized echo subtraction between the ultrashort and conventional multi-contrast images. Because the contrast mechanisms are concurrent, all image contrasts are self-registered. FIG. 4B: T2-weighting is encoded in the phaseof the background phase-corrected multi-contrast images. The vertical phase scale gives the values for the multi-contrast image phase in radians.

[0034] FIG. 5: Sagittal DREAMER musculoskeletal images for three anatomic locations (right knee, calf, and ankle / foot; 29 y.o. healthy female). In the knee, synovial fluid appears with low signal intensity on Tiw images and high intensity on T2W images (arrows; ψT1= 117°, ψT2= 159°). DREAMER adds additional radiological value for musculoskeletal MRI protocols because it also acquires bone-selective images, potentially obviating the need for additional CT. Additional non-bone short-T2 structures, like the patellar (arrow) or Achilles tendon (arrow), appear brightly in the normalized echo subtraction.

[0035] FIG. 6: Comparison of DREAMER image contrasts with their corresponding clinical standards for a pediatric patient (17 y.o. male; 6 min. scan). Bone-selective images are compared to CT, Tiw DREAMER images with Tiw MPRAGE, and T2W DREAMER images with T2W multi-slice FSEψT1= 114°, ψT2= 159°). The same anatomical features are present in DREAMER images and their clinically standard counterparts (e.g. skull, brain sulci and gyri, and cerebrospinal fluid as arrows). In a conventional head MRI protocol, Tiw MPRAGE and T2W FSE scans are separately acquired, prolonging the total examination time. The additional acquisition of a CT image requires a separate examination, which increases patient wait time and complicates the clinical workflow.

[0036] FIG. 7: Axial, sagittal, and coronal soft tissue DREAMER images of the knee (25 y.o. healthy male) for a sequence of phase modulation values i. In addition to static Tiw and T2W images, DREAMER enables dynamic adjustment of soft tissue contrast at reconstruction time. Instead of considering each image contrast in isolation, each contrast provides contextual information because i can be continuously modulated. This feature enhances the sensitivity of the sequence to differences in tissue T1 and T2 even when a particular tissue compartment occupies a small volume. As an example, synovial fluid, located in the small joint spaces between bones, appears dark on Tiw images and bright on T2W images. User-driven adjustment of i highlights the transition of fluid signal from dark to brightintensity as the images lose their T 1 -weighting and attain T2-weighting, making the fluid easy to detect (boxes).

[0037] FIG. 8: DREAMER image quality is robust to substantial reductions in scan duration. In MRI, a reduction in scan time is commensurate with a reduction in acquired data. An axial slice of the brain and skull (29 y.o. healthy female) is shown for the four fundamental DREAMER image contrasts (multicontrast magnitude, Tiw, T2W, and bone-selective; ψT1= 114°, ψT2= 153°). Separate scans were performed with durations of 1, 2, 3, 4, and 6 minutes. Differences in image quality between the scans of varying duration can mostly be appreciated upon close inspection of a small region of interest (box at upper left side of figure, shown in 3X Zoom at right side of figure). As the scan duration decreases, the images undergo increased blurring and image details are obscured. However, this loss in image quality is minimal with respect to the 6-minute benchmark, even for scan durations of 2-3 minutes.

[0038] FIG. 9: DURANDE (dual-RF, dual-echo UTE) and DREAMER bone-selective images validated with CT for a 17-year-old female pediatric patient with fibrous dysplasia and no history of prior cranial vault surgery. DURANDE and DREAMER share the same contrast mechanism for bone-selective imaging, and hence, full 3D skull renderings from the MR bone-selective images can be generated from the same U-Net model previously trained only on DURANDE images.

[0039] FIGS. 10A to 10F: Extracted image contrasts for various phase modulation values (0° ≤ ψ ≤ 170°, in increments of 10°). Contour plots depict the resulting signal magnitude as a function of species T1 and T2, and common tissues are plotted as points according to their T1 and T2 values at 3 Tesla1, 2. For the sake of simulation, proton density was assumed to be identical for all simulated tissue species. Sequence parameters were identical to the conventional TE acquisition used for DREAMER in this study (Table 1).

[0040] FIGS. 11A and 11B: Simulations examining the effect of off-resonance on the PSGRE steady-state signal magnitude and phase. FIG. 11 A: Plot of PSGRE signal magnitude vs. off-resonance for five T2 values. The off-resonant behavior of a species does not affect the PSGRE signal magnitude (barring R2’ effects). FIG. 11 B: Plot of PSGRE signal phase vs. off-resonancefor three T2 values. The steady-state signal phase shift varies linearly with the off-resonance of the species for both positive and negative steady-states (solid and dashed lines). Computation of the background phase calculates the phase shift caused by off-resonance (alternating dashed-dotted lines). The phase of the multi-contrast signal does not exhibit dependence on the off-resonance (dotted lines).

[0041] FIG. 12: Axial, sagittal, and coronal slices of a DREAMER acquisition at the right knee (25 y.o. healthy male). Synovial fluid, which appears dark on Tiw imaging and bright on T2W imaging (ψT1= 117°, ψT2= 168°), can be visualized in the joint space between the femur and the tibia (arrows). Other short-T2 signals, like the quadriceps and patellar tendons (arrows) are clearly visualized in the bone-selective image.

[0042] FIG. 13: Axial, sagittal, and coronal slices of a DREAMER acquisition at the right calf (25 y.o. healthy male; ψT1= 117°, ψT2= 168°). While subcutaneous fat (arrow) appears with high intensity in both Tiw and T2W imaging, muscle (arrow) appears with high intensity in Tiw images and lower intensity in T2W images. In bone-selective imaging, the tibia (arrow) and fibula (arrow) cortical bone can be clearly visualized.

[0043] FIG. 14: Axial, sagittal, and coronal slices of a DREAMER acquisition at the right ankle and foot (24 y.o. healthy female). Fluid in the joint spaces between bones appears dark on Tiw images and bright on T2W images (arrows; ψT1= 117°, ψT2= 159°). Articulations between individual bones, such as between the talus and the tibia (arrows), are clearly depicted.

[0044] FIG. 15: Sagittal DREAMER and FSE Tiw and T2W images of the right knee and ankle / foot (29 y.o. healthy female; ψT1= 54°, ψT2= 168°). Synovial fluid, located in the joint spaces of articulating bones, appears bright on T2W images and dark on Tiw images (knee, arrows; ankle, arrows). DREAMER soft tissue contrast can be made to match the Tiw and T2W contrasts of FSE acquisitions.

[0045] FIGS. 16A to 16C: Bloch simulations of a partially spoiled gradient echo sequence (PSGRE). FIG. 16A: Simulated plot of steady-state signal phase vs. T2 for various RF phase increments. Commonly used RF-spoiled gradient echo (“SPGR” or “FLASH”; 6 = 117°) and steady-state freeprecession (“FISP” or “GRASS”; 0 = 0°) sequences are plotted for reference. The vertical gray dotted line indicates the simulated TR. For tissues with T2 > TR, a small, nonzero RF phase increment produces a signal phase which is a monotonic function of tissue T2. The signal phases of conventional GRE sequences (SPGR / FLASH and FISP / GRASS) do not encode T2. Flipping the sign of the small RF phase increment flips the sign of the steady-state signal phase. FIG. 16B: Tiw images (i.e. signal magnitude decreasing with increasing tissue T1) are encoded in the signal for a small interval of phase modulation values (~0-20°). This Tiw signal is extracted by projecting the complex-valued signal onto the real axis after a phase modulation i. FIG.

[0046] 16C: Likewise, T2W images (i.e. signal magnitude increasing with increasing tissue T2) are encoded for a separate interval of phase modulation values (-90-170°).

[0047] FIGS. 17A to 17C: Bloch simulations of dual-echo UTE sequences for various RF phase increments. Note that T2 is plotted on a logarithmic scale and the vertical dotted gray line indicates the simulated TR. The application of RF pulses is assumed to be instantaneous (i.e. no loss in excitation efficiency when applied to short-T2 species). FIG. 17A: For tissues with short T2 (T2 < TR; e.g. bone), the steady-state signal magnitude is independent of the RF phase increment. Soft tissues with long T2 (T2 > TR; e.g. white / gray matter, cerebrospinal fluid, fat) accumulate phase based on their T2 value and thus exhibit a signal magnitude dependency on the RF phase increment. FIG.

[0048] 17B: An echo subtraction isolates short-T2 signals but may leave some long-T2 signal unsuppressed. FIG. 17C: A normalized echo subtraction further enhances short-T2 signal while suppressing long-T2 signal. Furthermore, a normalized echo subtraction exhibits no signal dependency on the RF phase increment.

[0049] FIG. 18 is a block diagram illustrating an exemplary system for controlling an MRI scanner.

[0050] DETAILED DESCRIPTION

[0051] Computed tomography (CT) and magnetic resonance imaging (MRI) acquire complementary structural information on the solid and soft tissues,respectively. Their joint use incurs numerous economic and logistical costs to the patient and imaging center. The subject matter described herein pertains in some embodiments to a new type of MRI scan, DREAMER (Dual Repetition and Echo Acquisition with Multi-contrast Encoding and Reconstruction), which rapidly and simultaneously acquires solid and soft tissue images, potentially replacing bone-selective CT images and sequentially acquired Ti- and T2-weighted soft tissue MR images. Advances in solid-state MRI and theoretical insights in phase-based T2 encoding produce a multi-contrast signal model that enables retrospective, user-driven customization of image contrast weighting. The model guides the development of an iterative image reconstruction algorithm for fast and high-resolution structural imaging of all tissue compartments. DREAMER images are self-registered and identical in resolution without additional post-processing. Moreover, DREAMER uses solely the hardware available on standard clinical MRI scanners. DREAMER may consolidate CT and MRI demand, decrease patient examination and wait times, reduce exposure to ionizing radiation, and simplify the radiological workflow.

[0052] In some embodiments, the presently disclosed subject matter provides a combined MRI method to simultaneously acquire bone-selective, T1-weighted, and T2-weighted contrasts in the same scan to produce a set of images that visualize the solid- and soft-tissue structure. The acquisition process can be greatly undersampled, hence reducing MRI scan times. The presently disclosed subject matter provides for rapidly acquiring images of all tissue compartments, thereby obviating the need for a lengthy MRI protocol or additional CT (a completely separate scan session). This combined capability for simultaneous and rapid imaging is useful for pediatric imaging, as well as imaging of other subjects. Ionizing radiation poses a great risk to pediatric patients receiving scans, and pediatric patients are typically less compliant if MR imaging protocols are too long.

[0053] Thus, in some embodiments, the presently disclosed subject matter pertains to methods for simultaneous, non-radiative imaging of soft and solid tissues with high contrast by magnetic resonance. In some embodiments, short-T2 selective (bone-selective or pseudo-CT), T1 -weighted, and T2-weighted images are jointly encoded and acquired by a dual-echo ultrashort echo pulse sequence, which implements a radiofrequency pulse schedule of small RF phase increment such that the T2-weighting of the imaged species is encoded along the measured signal (magnetization) phase. Because all the images can be simultaneously acquired in a single, rapid scan, all images are self-registered, obviating the need for additional registration during image post-processing. The technique can substantially abbreviate clinical magnetic resonance imaging protocols and eliminate the need for CT scanning. Also, the presently disclosed subject matter can reduce the use of unsafe ionizing radiation for imaging patients, lessen the burden and demand for limited clinical resources, and simplify the clinical workflow for patients, clinicians, and technicians.

[0054] In some embodiments, the presently disclosed subject matter provides methods for concurrent solid and soft-tissue imaging using a magnetic resonance imaging (MRI) system. An exemplary method comprises configuring an MRI pulse sequence to perform gradient-echo (GRE) imaging, wherein the pulse sequence is designed with a repetition time (TR) in the range suitable for distinguishing between short and long transverse relaxation times (T2); encoding long-T2 signals to obtain a spectrum of images, including T1 -weighted (T1w), T2-weighted (T2w), and other contrasts, by manipulating the phase schedule of the RF pulse train; acquiring signal data at multiple echo times, including an ultrashort echo time (UTE) and a longer, conventional echo time (CTE), to capture signals from both short-T2 and long-T2 tissues; and generating a bone-selective image by processing the acquired signal data to enhance short-T2 signals and suppress long-T2 signals, enabling differentiation of solid tissues such as bone. In some embodiments, multichannel receiver coils and advanced image reconstruction techniques are leveraged to reduce scan duration and improve image quality. In some embodiments, the acquired data is reconstructed into self-registered images representing both bone-selective and multi-contrast soft tissue information. In some embodiments, the presently disclosed subject matter provides comprehensive, simultaneous solid and soft tissue imaging without the use of ionizing radiation, reducing the need for a CT scan and separate MRI scans.The presently disclosed subject matter thus provides a method for magnetic resonance (MR) imaging. In some embodiments, the method comprises: during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject; receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject; receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject; generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals; generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times; repeating the above-listed steps for a plurality of repetition times that form the MR imaging scan; and simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and / or solid tissues.

[0055] The presently disclosed subject matter also provides a system for controlling a magnetic resonance imaging (MRI) scanner, the system comprising: at least one processor and a memory; and an MR imaging controller implemented by the at least one processor for: during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject; receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject; receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject;generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals; generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times; repeating the above-listed steps for a plurality of repetition times that form the MR imaging scan; and simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and / or solid tissues.

[0056] In some embodiments, the presently disclosed subject matter provides a non-transitory computer readable medium having stored thereon executable instructions that when executed by a processor of a computer control the computer to perform steps for MRI comprising: during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject; receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject; receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject; generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals; generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times; repeating the above-listed steps for a plurality of repetition times that form the MR imaging scan; and simultaneously generating, from first and second echo signals received during each repetition time of MRimaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1-weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and / or solid tissues.

[0057] In some embodiments, the RF pulses follow an RF phase schedule < >(n) determined by the RF phase increment 0, where n is an index corresponding to the repetition time and 0 is small but nonzero. In some embodiments, 0 is greater than 0 degrees to less than about 5 degrees, optionally ranging from about.5° to about 3°. In some embodiments, the RF phase schedule follows the formula φ(n)=φ(n−1)+nθ, where φ(n) is the phase of the RF pulse for the nth repetition time and φ(n − 1) is the phase of the RF pulse for the (n-1)th repetition time.

[0058] In some embodiments, receiving the first echo signal includes receiving the first echo signal at a time within a range of about 10 μs to about 100 μs after transmission of the RF pulse and receiving the second echo signal includes receiving the second echo signal at a time within a range of about 1 ms to about 4 ms after transmission of the RF pulse.

[0059] In some embodiments, simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images comprises applying phase modulation, optionally according to a phase modulation angle i according to the following equation:

[0060] x^ = |$R(xel^)|,

[0061] wherein

[0062]

[0063] denotes the extracted contrast, which was encoded along an angle i measured from the real axis. ℜ(u) denotes the voxel-wise real value of an arbitrary complex-valued image u, and |u| is the voxel-wise absolute value of u.

[0064] In some embodiments, within a single “repetition time”, there is a unit including an RF pulse, two echoes with encoding gradients for measuring thesignal, and a spoiler gradient for dephasing the residual magnetization (FIG.

[0065] 2A). This unit is the building block for a pulse sequence, and it is repeated (thousands of times) to create an “acquisition” (or “pass”). The duration of this repeated unit is denoted as “TR”, measured from the center of one RF pulse to the next. In some embodiments, for a single acquisition, the RF pulse phase of each unit (each TR) is calculated by Equation (1), forming an “RF phase schedule” < >(n) from the “RF phase increment” 9.

[0066] In some embodiments, in a “positive acquisition” 0 is used as the RF phase increment in Equation (1). In a negative acquisition, -0 is used (equal but opposite in sign) as the RF phase increment. In some embodiments, to measure and correct for the background phase, DREAMER performs both positive and negative acquisitions.

[0067] Thus, in some embodiments, a background phase is estimated with two acquisitions (or “passes”), one with a positive RF phase increment and one with equal but negative RF phase increment. In some embodiments, images of the MR are corrected with the background phase. Another approach is to use the background phase from a prior scan (e.g. a calibration scan). From the two acquisitions, the background phase can be estimated, but if the background phase is already known, one of the acquisitions can be eliminated.

[0068] T1 and T2 are intrinsic physical properties of materials (like how materials have a density, melting point, or specific heat). “Weighted” images are images where the signal intensity varies depending on physical parameters of the materials (e.g. biological tissues tend to have different T1s and T2s, and so they appear contrasted in weighted images). “T1 -weighted” images are images where the signal magnitude decreases for increasing material T1 (see FIG. 16B). As an example, T1 -weighted images of the brain tend to have bright white matter, darker gray matter, and dark fluid. “T2-weighted” images are images where the signal magnitude increases for increasing material T2 (see FIG. 16C). As an example, T2-weighted images of the brain tend to have bright fluid, darker gray matter, and dark white matter. “Short-T2 selective” images highlight materials where the T2 is less than the repetition time (TR) of the sequence. Everything else appears dark. Examplesof short-T2 materials are bone and tendon. In some embodiments, the T1-weighted images of the MR imaging subject pertain to soft tissue; T2-weighted images of the MR imaging subject pertain to soft tissue and / or bone tissue; and a short T2-selective image of the MR imaging subject pertains to bone tissue.

[0069] In some embodiments, the MR imaging subject is a pediatric subject. However, the scanner subject need not be a pediatric human subject. The subject can be an adult human subject, a non-human animal (adult or juvenile), or an inanimate object. Anything that can be scanned with an MRI can be scanned with this technique. Indeed, any suitable subject as would be apparent to one of ordinary skill in the art upon a review of the instant disclosure falls within the scope of the presently disclosed subject matter.

[0070] According to another aspect of the method, system, and / or non-transitory computer readable medium described herein, the MR images comprise accelerated, high-resolution, and / or self-registered MR images.

[0071] The presently disclosed subject matter also provides a magnetic resonance imaging (MRI) scanner. In some embodiments, the MRI scanner comprises: at least one magnet creating a static magnetic field; a RF amplifier and transmit coil configured to transmit an RF pulse with a phase into an MR imaging subject during a repetition time in an MR imaging scan; a receiving coil configured to receive, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject and to receive, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject; a gradient system configured for generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals and for generating and transmitting into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times; and a computer programmed to simultaneously generate, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1 -weighted images of the MR imaging subject with different contrast levels, T2-weighted imagesof the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1 -weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and short T2-selective images of the MR imaging subject of soft and / or solid tissues.

[0072] In some embodiments, the RF amplifier and transmit coil is configured to transmit RF pulses following an RF phase schedule < >(n) determined by the RF phase increment 0, where n is an index corresponding to the repetition time and 0 is small but nonzero. In some embodiments, 0 is greater than 0 degrees to less than about 5 degrees, optionally ranging from about.5° to about 3°. In some embodiments, the RF phase schedule follows the formula (n)= (n-1)+n0, where < >(n) is the phase of the RF pulse for the nth repetition time and (p(n - 1) is the phase of the RF pulse for the (n-1)th repetition time.

[0073] In some embodiments, receiving the first echo signal includes receiving the first echo signal at a time within a range of about 10 μs to about 100 μs after transmission of the RF pulse and receiving the second echo signal includes receiving the second echo signal at a time within a range of about 1 ms to about 4 ms after transmission of the RF pulse.

[0074] In some embodiments, simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images comprises applying phase modulation, optionally according to a phase modulation angle i according to the following equation:

[0075]

[0076] = |5R(xe^)|,

[0077] wherein

[0078]

[0079] denotes the extracted contrast, which was encoded along an angle i measured from the real axis. ℜ(u) denotes the voxel-wise real value of an arbitrary complex-valued image u, and |u| is the voxel-wise absolute value of u.

[0080] In some embodiments, within a single “repetition time”, there is a unit including an RF pulse, two echoes with encoding gradients for measuring the signal, and a spoiler gradient for dephasing the residual magnetization (FIG.

[0081] 2A). This unit is the building block for a pulse sequence, and it is repeated(thousands of times) to create an “acquisition” (or “pass”). The duration of this repeated unit is denoted as “TR”, measured from the center of one RF pulse to the next. In some embodiments, for a single acquisition, the RF pulse phase of each unit (each TR) is calculated by Equation (1), forming an “RF phase schedule” < >(n) from the “RF phase increment” 9.

[0082] In some embodiments, in a “positive acquisition” 0 is used as the RF phase increment in Equation (1). In a negative acquisition, -0 is used (equal but opposite in sign) as the RF phase increment. In some embodiments, to measure and correct for the background phase, DREAMER performs both positive and negative acquisitions.

[0083] Thus, in some embodiments, the computer is configured to estimate a background phase with two acquisitions (or “passes”), one with a positive RF phase increment and one with equal but negative RF phase increment. In some embodiments, the computer is configured to correct images of the MR with the background phase. Another approach is to use the background phase from a prior scan (e.g. a calibration scan). From the two acquisitions, the background phase can be estimated, but if the background phase is already known, one of the acquisitions can be eliminated.

[0084] T1 and T2 are intrinsic physical properties of materials (like how materials have a density, melting point, or specific heat). “Weighted” images are images where the signal intensity varies depending on physical parameters of the materials (e.g. biological tissues tend to have different T1s and T2s, and so they appear contrasted in weighted images). “T1 -weighted” images are images where the signal magnitude decreases for increasing material T1 (see FIG. 16B). As an example, T1 -weighted images of the brain tend to have bright white matter, darker gray matter, and dark fluid. “T2-weighted” images are images where the signal magnitude increases for increasing material T2 (see FIG. 16C). As an example, T2-weighted images of the brain tend to have bright fluid, darker gray matter, and dark white matter. “Short-T2 selective” images highlight materials where the T2 is less than the repetition time (TR) of the sequence. All other materials in a short-T2 selective image appear dark or with suppressed intensity. Examples of short-T2 materials are bone and tendon. In some embodiments the T1 -weightedimages of the MR imaging subject pertain to soft tissue; T2-weighted images of the MR imaging subject pertain to soft tissue and / or bone tissue; and a short T2-selective image of the MR imaging subject pertains to bone tissue.

[0085] In some embodiments, the MRI scanner is configured to perform for a plurality of repetition times to form a MR scan.

[0086] In some embodiments of the presently disclosed subject matter, an unbalanced spoiler gradient is employed at the end of each repetition time.

[0087] In some embodiments of the presently disclosed subject matter, acquisition passes are employed to estimate the background phase (one with positive RF phase increment, one with equal but negative RF phase increment). Alternatively, a calibration scan can be used to estimate the background phase and / or a physics-constrained signal separation of image and background phase can be performed at processing; in this case, a second pass is not needed.

[0088] In some embodiments of the presently disclosed subject matter, image reconstruction and postprocessing can include reconstruction of each dataset to produce phase-corrupted images for each dataset. Reconstruction can be by conventional Fourier or non-uniform Fourier with gridding and / or can also be model-based (iterative methods) or by kernel methods or by deep learning. Reconstruction can also include regularization.

[0089] In some embodiments, the presently disclosed subject matter includes estimating the background phase from the image phases of the acquisition pairs (positive and negative). In some embodiments, removal of the background phase is by phase subtraction.

[0090] In some embodiments, the presently disclosed subject matter includes estimating the multi-contrast image by summing the phase-corrected images from each acquisition (or any linear combination of the phase-corrected images). In some embodiments, the presently disclosed subject matter includes extraction of image contrasts by phase rotation of the complexvalued multi-contrast image, then projection onto the real axis. In general, any selective function of signal phase can be used (does not have to be projection). In some embodiments, the presently disclosed subject matter includes creating a retrospective contrast adjustment method (“dial” feature)by extracting a sequence of contrasts for incremental changes in phase rotation. Psi ( >) is a parameter that is used in the “dialing” feature. By way of elaboration, the phase modulation parameter i is a parameter for processing a signal, used in a calculation after the signal has been measured and recorded. Equation (2) set forth elsewhere herein details the computational step for processing the collected multi-contrast signal x. The output of the processing step is a single image contrast (denoted

[0091]

[0092] and different choices of psi compute different image contrasts. In some examples, Equation (2) is computed for various values of i, showing that the image contrast can be adjusted by adjusting the value of i. Because this computation takes place after the scan, the contrast can be retrospectively adjusted (by adjusting i ) even after the scan is over. The feature of retrospective contrast adjustment is not typical of conventional MRI scans.

[0093] In some embodiments, the presently disclosed subject matter includes reconstruction of a bone-selective image from multi-contrast images derived from ultrashort and conventional echo datasets, including standard echo subtraction; rescaled echo subtraction; and / or normalized subtraction -direct OR iterative (e.g. weighted least-squares).

[0094] In some embodiments, the presently disclosed subject matter includes a computer, and / or at least one processor and a memory and an MR imaging controller implemented by the at least one processor, for controlling timing of RF pulses and gradient pulses; receiving and storing the raw MR signal; performing image reconstruction and postprocessing; and / or displaying MR images.

[0095] While the following terms are believed to be well understood by one of ordinary skill in the art, the following definitions are set forth to facilitate explanation of the presently disclosed subject matter.

[0096] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood to one of ordinary skill in the art to which the presently disclosed subject matter belongs.

[0097] Following long-standing patent law convention, the terms “a”, “an”, and “the” refer to “one or more” when used in this application, including the claims.

[0098] The term “and / or” when used in describing two or more items or conditions, refers to situations where all named items or conditions arepresent or applicable, or to situations wherein only one (or less than all) of the items or conditions is present or applicable.

[0099] The use of the term “or” in the claims is used to mean “and / or” unless explicitly indicated to refer to alternatives only or the alternatives are mutually exclusive, although the disclosure supports a definition that refers to only alternatives and “and / or.” As used herein “another” can mean at least a second or more.

[0100] Unless otherwise indicated, all numbers expressing quantities of time, concentration, dosage and so forth used in the specification and claims are to be understood as being modified in all instances by the term “about”. Accordingly, unless indicated to the contrary, the numerical parameters set forth in this specification and attached claims are approximations that can vary depending upon the desired properties sought to be obtained by the presently disclosed subject matter.

[0101] As used herein, the term “about”, when referring to a value is meant to encompass variations of in one example ±20% or ±10%, in another example ±5%, in another example ±1%, and in still another example ±0.1% from the specified amount, as such variations are appropriate to perform the disclosed methods.

[0102] As used herein, ranges can be expressed as from “about” one particular value, and / or to “about” another particular value. It is also understood that there are a number of values disclosed herein, and that each value is also herein disclosed as “about” that particular value in addition to the value itself. For example, if the value “10” is disclosed, then “about 10” is also disclosed. It is also understood that each unit between two particular units are also disclosed. For example, if 10 and 15 are disclosed, then 11, 12, 13, and 14 are also disclosed.

[0103] 1. Introduction

[0104] Computed tomography (CT) and magnetic resonance imaging (MRI) are diagnostic mainstays of modem radiology. About 80 million CT scans and 40 million MRI scans are annually performed in the United States1’2Furthermore, these numbers are expected to increase at a rate of 1-5% per year3’4

[0105] CT and MRI differ in how images are acquired and thereby produce complementary tissue information. In CT, X-rays are emitted from a source, attenuated as they pass through the body, and measured by an opposing detector array. The measured X-ray intensities are sensitive to differences in tissue attenuation. Highly attenuating “solid” tissues, like densely calcified bone, are depicted with high signal intensity and excellent contrast in a CT image5.

[0106] MRI examinations are typically comprised of several scans, which jointly aim to determine the underlying pathology. Each individual scan consists of a single pulse sequence, a collection of radiofrequency (RF) pulses and magnetic gradients with pre-specified timing, and each pulse sequence sensitizes the resulting image to a targeted set of physical properties indicative of soft tissue pathology6. Because pulse sequences can be flexibly and arbitrarily designed, a virtually limitless number of MRI contrasts may be obtained7. Clinical practice most commonly obtains images weighted by two fundamental tissue properties, Ti and T2, corresponding to the relaxation rates of the longitudinal and transverse magnetization. These two contrasts are used for visualizing soft tissue structure and identifying fluid, respectively8.

[0107] CT and MRI have different limitations. For example, CT scans are rapid (~30 s), but MRI examinations are comparatively slow. Each MRI scan, which typically achieves a single image contrast, has a duration of a few minutes. Because multiple MRI contrasts are needed to jointly assess structure, function, and pathology, MRI examinations are much longer than CT scans9. Moreover, unlike MRI, CT uses ionizing radiation and increases the risk of cancer. This risk is higher for children, who are more sensitive to ionizing radiation than adults10.

[0108] Certain clinical presentations may benefit from ordering both CT and MRI examinations. Examples include craniofacial anomalies affecting both the bone and the underlying neurology11-13; bone metastases, which may involve both the cortical bone and marrow cavity14; pediatric trauma, where CT and MRI are used to visualize skeletal and parenchymal injuries15’16; and lowerextremity stress fractures of the bone, where symptoms may mimic injuries of the muscle or tendon17. Ideally, these scenarios would use CT to assess the bony anatomy and MRI to produce exquisite soft tissue contrast.

[0109] In total, the two-modality scenario creates a multifaceted economic and logistical burden for the patient and imaging center. Patients needing both MRI and CT undergo longer total exam and wait times and experience additional scheduling complexities as opposed to patients needing only CT18’19Two sets of personnel are needed to operate the scanners and interpret the images19. The radiation used in a CT scan also undercuts the safety benefits of an MRI. Finally, because CT and MRI do not typically acquire at the same resolution, images would need to be rescaled and co-registered so that they could be jointly viewed for diagnosis or planning20, introducing additional workflow complexities.

[0110] A desirable imaging protocol would combine desirable properties of CT (bone-selective, rapid, high-resolution) and MRI (enhanced and multiple soft-tissue contrast, no ionizing radiation) in a single 3D scan without the need for additional investment in clinical infrastructure. Such a method would produce an efficient mode of scanning that maintains patient safety, simplifies the clinical workflow, and consolidates demand for high-resolution clinical imaging.

[0111] Existing techniques attempt to satisfy these requirements but fall short. Phase-contrast and photon-counting CT may enhance soft tissue contrast relative to conventional CT. However, they require sophisticated hardware, have so far seen limited clinical use, and use ionizing radiation21’22“Blackbone MRI” techniques use gradient echo (GRE) sequences with small RF flip angles to acquire a 3D image of the soft tissue. The resulting proton-density weighted (PDw) images have relatively uniform soft tissue signal, but as with all conventional MRI methods, signal from bone water, water hydrogen-bonded to the collagen matrix and the major source of detectable protons in bone23, decays before it can be measured. This is because bone water T2, the time constant with which bone water signal decays after RF excitation, is several times shorter than the echo time (TE) used for signal measurement24. For black-bone techniques, identification of bone is based on the absence ofsignal, which conflates bone with air25’26Furthermore, PDw images are not as commonly used as Ti-weighted or T2-weighted (Tiw and T2W) images, which would have to be acquired in succession, prolonging examinations. Finally, neural network-based synthetic methods have been developed to generate CT from MRI27-30, but the training requires large datasets to cover the wide variability of patient phenotypes, demographics, and pathologies. Even when large datasets are available, issues regarding generalizability and out-of-distribution samples may occur31 32. Hence, medical image synthesis is costly and challenging.

[0112] The presently disclosed subject matter provides a new MRI pulse sequence (DREAMER: Dual Repetition and Echo Acquisition with Multicontrast Encoding and Reconstruction) to perform simultaneous solid (bone-selective) and multi-contrast soft tissue (Tiw and T2W) imaging in a single scan. GRE imaging, which is conventionally T1 -weighted33, can be combined with technical advancements in solid-state MRI34’35and theoretical insights on phase-based T2 encoding36’37to produce an imaging technique with all the desirable properties of CT and MRI. In a GRE, RF pulses are regularly spaced in time (“repetition time”, or TR). The main technical claim of this work is that in short-TR GRE sequences (4-8 ms TR), the signal behavior of any tissue species depends primarily on whether it has a short or long T2 (T2 TR, respectively). This categorization describes how the magnetization of each species interacts with the RF pulse train of a GRE. The signal of short-T2 species, like bone water, decays within a single TR. Sampling the signal at an ultrashort TE (UTE; immediately after the RF pulse) and a conventional TE (CTE; TE 4-1 Ox the T2 of collagen-bound water in bone) probes this decay. Because long-T2 signal, characteristic of soft tissue, minimally decays between the two TEs, a subtraction of the two TE images retains short-T2 signal and suppresses long-T2 signal38 39to produce a bone-selective image. By comparison, long-T2 components decay over the course of multiple TRs. Therefore, these components can be encoded along the phase of the multicontrast MRI signal by manipulating the phase schedule of the train of RF pulses36’37. Phase modulation of the multi-contrast signal data providesretrospectively adjustable and customizable Tiw and T2W contrast for long-T2 species, thereby enhancing visualization of soft-tissue structures.

[0113] In principle, the proposed pulse sequence for concurrent solid- and soft-tissue imaging may be used with current clinical MRI scanners and receiver coils, avoiding the need for investment in new hardware by imaging centers. Because the multi-contrast encoding is based on MRI physics and is not synthetic or generative, it obviates the accumulation of large training datasets and confrontation of out-of-distribution data. Furthermore, the method leverages the availability of multi-channel receiver coils40to substantially reduce the scan duration41. The new scanning method is compatible with model-based image reconstruction techniques42’43, enabling fast and concurrent solid- and soft-tissue imaging. 3D bone-selective, Tiw, and T2W images are self-registered, high-resolution, and spatially isotropic, potentially producing a one-stop shop for structural solid and soft tissue imaging.

[0114] 2. Results

[0115] 2.1 Simultaneous Tiw and T2W contrast mechanism

[0116] Classically, weighted imaging by T1 or T2, the time constants for longitudinal recovery and transverse decay of magnetization, is achieved by adjusting the acquisition TE and TR44. In sequences with short TR and short TE (TE < TR < T1), tissue species with shorter T1 (e.g. white matter, subcutaneous fat, and bone marrow) produce a larger signal intensity because the longitudinal magnetization recovers faster than that of species with longer T1. Similarly, in sequences with long TR and long TE (T2 - TE < TR), species with longer T2 (e.g. cerebrospinal fluid, subcutaneous fat, bone marrow, and synovial fluid) produce a larger intensity because the transverse magnetization decays slower than that of species with shorter T2. In each acquisition, the spatial distribution of T1 or T2 weights the relative signal magnitudes of different tissues to produce a unique image contrast.

[0117] Unlike classical methods, a partially spoiled gradient echo (PSGRE) simultaneously achieves T1- and T2-weighting. The image processing workflow for a PSGRE is shown in FIG. 1 A. In addition to a short TR and short TE to classically achieve T1 -weighting, a PSGRE uses a small, quadraticallyincreasing RF phase increment so as to also encode T2-weighting in the signal phase (i.e. the signal phase increases for increasing tissue T236). When T1-and T2-weighting are encoded along different phase values in the multicontrast signal, they may also be decoded by taking solely the signal component along those same phase values, thereby extracting the desired signal weighting and achieving a specific image contrast.

[0118] Because the background phase arbitrarily changes the signal phase, it must be removed to correctly extract image contrast. Hence, a PSGRE acquisition typically acquires two steady states (positive and negative)36’37. From these, the background phase and image phase are estimated and corrected to produce a multi-contrast image, which represents the underlying superposition of magnetization components. Notably, in FIG. 1A, the image phase of the multi-contrast image is observed to encode T2 (e.g. higher signal phase for cerebrospinal fluid, a long-T2 structure). Contrast extraction recovers individual Tiw and T2W contrasts by phase modulation of the multicontrast image. Because the acquisition is simultaneous, all images are selfregistered.

[0119] 2.2 Bone-selective contrast mechanism

[0120] FIG. 1B outlines image processing steps for a dual-echo UTE sequence which images short-T2 tissues (e.g. bone, tendons, ligaments, epidermis, and hair). In a dual-echo UTE, each pulse is followed by two echo readouts. The ultrashort echo, acquired immediately after the RF pulse, captures a nonzero short-T2 signal, while the conventional echo, acquired with a delay following the RF pulse, captures a second image where the short-T2 signal has fully decayed. A normalized echo subtraction between echo images enhances short-T2 (solid) tissues and suppresses long-T2 (soft) tissues. Because the echo acquisitions are interleaved, all three images are selfregistered. Typical dual-echo UTE sequences use full RF spoiling and short TRs, yielding pure Ti-weighting in each echo image.

[0121] 2.3 DREAMER: Concurrent bone-selective, Tiw and T2W imaging DREAMER is a pulse sequence which combines the contrast mechanisms of PSGRE and dual-echo UTE acquisitions for simultaneous solid and soft tissue imaging (FIG. 2A). A train of RF pulses with a PSGREphase schedule drives the simultaneous Ti- and T2-weighted contrast for long-T2 species. Two passes are performed to remove the background phase. The rapid short-T2 decay of bone water signal is captured by the dual-echo acquisition. Bone water signal is then selectively isolated by a normalized echo subtraction. In total, DREAMER produces four datasets (2 echoes x 2 passes; FIG. 2B), used to compute a bone-selective image and tunable Tiw and T2W image contrasts. All 3D images are self-registered and of isotropic resolution.

[0122] The four datasets in DREAMER are used to retrieve ultrashort and conventional echo multi-contrast images (FIG. 3A). In the two multi-contrast images, the T2-weighting of soft tissues is encoded in the image phase. The two images are used to derive all solid and soft tissue image contrasts by either normalized echo subtraction or contrast extraction (FIG. 3B). In summary, the DREAMER technique entails various tools to encode and decode solid and soft tissue images of multiple contrasts within a single scan.

[0123] 2.4 Imaging with DREAMER

[0124] Multi-contrast magnitude, Tiw, T2W, and bone-selective images are all acquired in a single DREAMER scan. These representative contrasts are shown for the head, knee, calf, and foot in FIGS. 4A, 4B, and 5. FIGS. 12-14 also show images for the knee, calf, and foot for all three principal anatomical planes. For each DREAMER scan, two multi-contrast images (UTE and CTE) form the basis for simultaneous Tiw, T2W, and bone-selective imaging. The image phase is a feature of the complex-valued multi-contrast images because it forms the basis for additional encoding of T2-weighting (FIG. 4B).

[0125] Because both multi-contrast images encode various weighted image contrasts in the signal phase, Tiw and T2W images can be extracted from either one. A short-T2 enhancing image is derived from a normalized echo subtraction between the ultrashort and conventional multi-contrast images.

[0126] Three contrasts commonly used for craniofacial imaging are included in a single DREAMER acquisition. A pediatric patient who had three years ago sustained trauma to the face was imaged with DREAMER, Tiw MPRAGE (“magnetization-prepared rapid gradient echo”), T2W FSE (“fast spin-echo”), and CT (FIG. 6). Here, Tiw, T2W, and bone-selective DREAMER contrasts arecompared to their respective clinical standards. The same anatomical features of the skull and brain are visible in both DREAMER images and the clinical standards. Tiw DREAMER achieves less white-gray matter contrast than Tiw MPRAGE because the latter uses an inversion-recovery technique to enhance Ti -weighting. However, inversion-recovery pulses reduce the scan efficiency of a sequence and prolong scan time. Moreover, DREAMER acquires solid and multiple soft tissue contrasts all at once, while Tiw MPRAGE and T2W FSE sequences need to be acquired separately. CT images require a separate examination and use ionizing radiation. Overall, DREAMER may be a viable alternative to standard craniofacial MRI and CT for diagnosis, surgical planning, and clinical follow-up.

[0127] Tiw and T2W images are often jointly used to detect and identify tissues. In musculoskeletal MRI, Tiw and T2W FSE images visualize synovial fluid, which occupies the small joint spaces and which, like other fluids, has a long T2. Like in Tiw and T2W FSE images, synovial fluid appears dark and bright in Tiw and T2W DREAMER images, respectively (FIG. 15). Even when provided with both contrasts, it may be difficult to locate the differences in signal intensity and identify synovial fluid.

[0128] The adjustable contrast feature of DREAMER images enhances sensitivity to patient-specific differences in anatomy and tissue composition. This feature is demonstrated in visualizing the synovial fluid in the knee (FIG.

[0129] 7). Starting from a Tiw image, as the phase modulation value is increased, images lose their Ti-weighting, eventually attaining T2-weighting. Synovial fluid is easily identified because it increases in signal intensity during this user-driven adjustment process. Because all the DREAMER images are selfregistered and the phase modulation parameter can be continuously adjusted, multiple DREAMER contrasts can be jointly considered to assess structure and pathology. Additional examples of contrast adjustment for several phase modulation values are provided in FIGS. 4A, 4B,, 5, 6, and 7; FIGS. 10A to 10F; and FIGS. 12-15, (head, knee, calf, and foot), in which various soft tissue structures are brought in and out of the viewer’s focus as they gain and lose signal intensity.The solid- and soft-tissue contrast mechanisms of DREAMER can be directly incorporated into a model-based reconstruction. Model-based reconstruction leverages improvements in receiver coil hardware and advances in imaging theory to achieve significant reductions in scan time41 43. The four fundamental DREAMER contrasts, derived from the two multicontrast images, can be reconstructed from scans of varying duration (FIG.

[0130] 8). As the scan duration decreases, blurring ensues and the apparent resolution decreases. However, this blurring is noticeable only at very high acceleration factors. Image quality is reasonably maintained even for short scans of 2-3 minutes. A short scan yielding several isotropic and co-registered solid and soft tissue images over a large three-dimensional field-of-view may increase the value proposition of MRI over CT.

[0131] Because the bone-selective contrast mechanism of DREAMER is identical to that of a dual-echo UTE sequence, previously developed image processing tools in the authors' laboratory for bone-selective imaging may be applied to DREAMER images without modification. Figure 9 compares bone-selective images and renderings, derived from CT, DURANDE

[0035] (i.e., a dual-echo UTE sequence), and DREAMER for a pediatric patient with a clinical history of fibrous dysplasia. The U-Net model developed and used in a previous study

[0034] to segment DURANDE bone images, was applied to DREAMER bone images without additional training (i.e., U-Net training data consisted of only DURANDE images, as in the previous study, and did not include any DREAMER data). A qualitative comparison of the anatomy depicted in the bone-selective images and 3D skull renderings suggests that sequence differences between DURANDE and DREAMER do not substantially and negatively impact the bone-selective contrast mechanism nor the segmentation model performance.

[0132] Reconstruction times increased with the amount of raw k-space data collected (i.e., the duration of the scan). By way of example and not limitation, timing benchmarks to compute each processing step or image in FIGS. 2A, 2B, 3A and 3B were as follows for 3- and 6-min DREAMER scans, respectively: coil sensitivity map estimation, 1.8 and 3.4 min; image reconstruction of the four phase-corrupted multi-contrast images with jointwavelet regularization, 104.7 and 134.4min; reconstruction of the bone-selective image, 0.3 and 0.3 min. The time calculating the background phase and extracting soft-tissue contrasts was negligible because these are element-wise array operations. As would be apparent to one of ordinary skill in the art upon a review of the instant disclosure, the representative, nonlimiting timing benchmarks above can vary based on the specification of the computer used, the receiver coil hardware chosen, and the like.

[0133] 3. Discussion

[0134] A new MRI pulse sequence, referred to as DREAMER, to simultaneously acquire solid and multiple-contrast soft tissue images is disclosed herein. DREAMER is compatible with existing clinical MRI scanners and receiver coils, and the resulting images are 3D, isotropic in resolution, and self-registered. Like conventional T1w and T2w MRI, T1w and T2w DREAMER images are sensitive to variations in tissue composition. Unlike conventional MRI, DREAMER multi-contrast images allow for retrospective and user-customizable weightings of T1 and T2, giving clinicians a potential tool to detect and diagnose pathology. Unlike axial CT, DREAMER images can be resliced along planes with arbitrary orientation without incurring a loss in resolution. DREAMER achieves similar soft tissue contrast to standard clinical MRI techniques (T1w MPRAGE and T1w and T2w FSE) and acquires solid-tissue structural information comparable to CT. Advanced reconstruction techniques34’41-43are used with DREAMER to reduce scan time without substantially compromising image quality. In total, DREAMER is a radiation-free imaging technique that combines the advantages of conventional MRI and CT. When examinations from both modalities are warranted, a single DREAMER scan may simplify the clinical workflow, consolidate demand for imaging, and reduce economic burdens on the patient and imaging center.

[0135] In MRI, the data acquisition process is determined by the pulse sequence, which may be designed in a very flexible and general manner and implemented on any clinical MRI scanner. The presently disclosed subject matter devises a pulse sequence which concurrently encodes short- and long-T2 signal. DREAMER is constructed on the principle of a gradient echo (GRE),which implements a regularly spaced train of RF pulses to drive the magnetization towards steady state. In steady state, short- and long-T2 spins interact differently with the train of RF pulses. This observation informs encoding of short- and long-T2 signal by the dual-echo and RF phase schedule mechanisms, respectively. Subsequent decoding occurs by normalized echo subtraction and contrast extraction along the signal phase. Because the sequence simultaneously encodes short- and long-T2 signal, the images are self-registered and of identical resolution, simplifying downstream tasks in the clinical workflow. By using a multi-channel receiver coil and advanced image reconstruction, the acquisition time of all solid- and soft-tissue contrasts is reduced to 2-3 minutes.

[0136] Further approaches are also addressed. First, depending on the number of receiver coil channels and the image matrix size, the image reconstruction runtime is relatively lengthy (~1-5 hours). Algorithmic optimizations and GPU computing reduce the time needed to reconstruct all solid- and soft-tissue images45’46. Second, GRE or GRE-derived sequences with unbalanced gradient moments, like DREAMER, are sensitive to flow and off-resonance artifacts. In the head, where cerebrospinal fluid flows between the ventricles and the spinal cord, if the flow velocity is sufficiently fast, signal voids may appear where fluid is expected due to deviations in RF spoiling from the prescribed RF phase schedule47. In musculoskeletal applications where subcutaneous fat contributes a substantial amount of signal, lipid and water spins resonate with different frequencies and create an increasing phase shift along the readout, resulting in generalized blurring for radial acquisitions like DREAMER48. Methods to reduce or eliminate these undesirable artifacts are addressed. Third, DREAMER uses two passes to estimate and remove the background phase. A second pass renders DREAMER susceptible to patient motion between the two passes, which may result in an incorrect background phase estimation and an inability to correctly perform contrast extraction. If the background phase were already known, DREAMER scan time would be halved. Finally, in the present work data was acquired from a small number of healthy subjects and one patient (to compare CT and bone-selective images). Additional approaches aim to enroll patients with various clinical presentationsto further validate the efficacy of DREAMER for simultaneous solid and soft tissue imaging. Reader studies by radiologists are performed to assess the image quality and diagnostic value of DREAMER in a clinical setting.

[0137] Secondary considerations for additional approaches include increasing the scanning efficiency of DREAMER by optimizing the encoding trajectory and incorporating deep learning for image reconstruction. DREAMER, in an example embodiment, uses a center-out, ramp-up radial readout trajectory for both its ultrashort and conventional echo readouts. While this trajectory is simple to implement for ultrashort echo imaging, more sophisticated schemes49’50have been shown to acquire data more efficiently. Deep learning MRI reconstruction may enable faster scan acquisition by leveraging learned image priors32.

[0138] 4. Outlook

[0139] DREAMER benefits from the flexibility of MRI pulse sequence design to achieve joint encoding and decoding of solid and soft tissue signals. Specifically, DREAMER encodes the dichotomy between short- and long-T2 tissues in the two echoes and the spectrum of long-T2 species along the signal phase. The simultaneously acquired and self-registered bone-selective, T1w, and T2w images from DREAMER are a potential substitute for CT and individual T1w and T2w MRI scans. DREAMER images facilitate easy assessment of the same anatomical region across different image contrasts.

[0140] Clinically, DREAMER can create a viable alternative to CT, which is limited in soft tissue contrast and uses ionizing radiation, and conventional MRI, which does not typically acquire bone-selective images and performs an inefficient sequential acquisition of scans with differing tissue contrast. Because CT and MRI are ubiquitous in modern radiology, simultaneous, rapid, and high-resolution solid and multiple-contrast soft tissue imaging has clinical utility whenever the pathology involves both skeletal and soft-tissue anatomy. DREAMER can decrease examination and scheduling complexity, increase patient safety by obviating the need for ionizing radiation, and reduce logistical complexities and personnel costs when patients need CT and MRI. Because DREAMER is entirely encapsulated in the design of the pulse sequence andimage reconstruction, its benefits are attainable with standard clinical MRI scanners without the need for additional custom hardware. For imaging centers with high patient throughput, DREAMER can provide a straightforward method to consolidate MRI and CT examinations and streamline radiological workflows. For patients and clinicians, DREAMER offers a one-stop shop for structural solid and soft tissue imaging.

[0141] 5. Methods

[0142] 5.1 Partially RF-spoiled gradient echo: concurrent T1w and T2w imaging Because of their short TRs, GRE sequences always achieve some T1-weighting, provided that the RF flip angle is reasonably large. Whether an additional T2-weighting is achieved depends on the prescribed RF phase schedule

[0143]

[0144] which dictates how the residual magnetization at the end of each TR contributes to the signal in subsequent TR cycles (n denotes the RF pulse index). For the magnetization to achieve steady state, a single RF phase increment 0, with -180° < 6 < 180°, may be used to determine the phase schedule by the recurrence relation:

[0145] (p(n) = (p(n — 1) + nd. (1) For example, an RF-spoiled GRE (“SPGR” or “FLASH”), using an RF phase increment of 117°, achieves pure Ti-weighting because the selection of 6 causes the residual magnetization from previous TRs to accumulate incoherently33. An unbalanced steady-state free precession sequence (“FISP” or “GRASS”; 6 = 0° or, more generally, a phase schedule of < >(n) = nd) uses a train of RF pulses that is phase coherent and achieves an image weighting that is a function of both T1 and T251.

[0146] A partially RF-spoiled gradient echo (PSGRE) introduces a small amount of phase incoherence by reducing the RF phase increment to a small, nonzero value (e.g. 1-4°). In this small RF phase increment regime, tissues with longer T2 accumulate a larger phase (i.e. the measured signal phase monotonically increases with increasing tissue T2), thereby encoding an additional T2-weighting along the phase of the signal36(FIG. 16A). Bloch simulations were performed using Sycomore52The measured steady-state signal, which consists of a superposition of free induction decay, spin-echo,and stimulated-echo magnetization components, is denoted as the multicontrast signal or multi-contrast image because it stores both T1w and T2w image components in the signal along different values of the signal phase. This phenomenon forms the basis for simultaneous T1w and T2w imaging.

[0147] The selection of a specific image contrast is performed by rotation of the complex-valued multi-contrast image x by a phase modulation angle i before projection on the real axis:

[0148] x^, = |5R(xe^)|. (2) The operation (2) is called contrast extraction,

[0149]

[0150] denotes the extracted contrast, which was encoded along an angle i measured from the real axis.

[0151] 5R(u) denotes the voxel-wise real value of an arbitrary complex-valued image u, and | u| is the voxel-wise absolute value of u.

[0152] FIGS. 16B and 16C illustrate how contrast extraction can achieve T1-and T2-weighting. For the simulated PSGRE parameters, Ti-weighting (signal magnitude decreasing with increasing T1) is attained for a small interval of phase modulation values (0° < i < 20°). Likewise, T2-weighting (signal magnitude increasing with increasing T2) is achieved on a separate interval (90° < I / J < 170°). Because the phase modulation i is tunable at the point of reconstruction, T1- and T2-weighting may be retrospectively adjusted to suit the needs of the user. Notably, all unique image contrasts are extracted for phase modulation values of 0° < i < 180°. This is demonstrated by showing that is TT-periodic:

[0153] x

[0154]

[0155] ^+n= |9t(xetW,+7r:)) | = |3t(-xe'^)| = |-1 • 3t(xe'^)| = |5R(xe^)| = x^,. (3) Thus, by restricting phase modulation values to 0° < i < 180°, all soft tissue contrasts that may be reconstructed from the phase-based encoding technique of a PSGRE will be included. FIGS. 10A to 10F depict in vivo brain images from a PSGRE acquisition for varying phase modulation values used for contrast extraction. The corresponding contour plots were created by simulating the same PSGRE parameters used in the acquisition.

[0156] In practice, the steady-state magnetization is corrupted by a spatially varying background phase shift, caused by acquisition imperfections like inhomogeneities in the static field and RF excitation. The background phaseis modeled by a pointwise multiplicative term ei<p(-rwith <p = <p(r) serving as a gentle reminder that the background phase may assume a different value at each voxel36. The forward model for a phase-corrupted PSGRE acquisition is y = FSe^x. (4) S denotes pointwise multiplication by coil sensitivities, F encodes the Fourier sampling trajectory of the PSGRE sequence, and y is the acquired multi-coil k-space dataset. A naive reconstruction using coil sensitivities S would yield not x, but ei?,(r)x, a phase-corrupted version of the multi-contrast image. Without knowledge of the background phase, ei<p(r)cannot be separated from x. Because the image contrasts are stored in the phase of x, a spatially varying background phase disrupts the contrast extraction method described by (2).

[0157] To resolve the issue of the background phase, a two-pass strategy is implemented in which the second (negative) pass flips the sign of the RF phase increment used in the first (positive) pass36 37. FIG. 16A shows that flipping the sign of 0 also flips the sign of the steady-state signal phase. As a result, the signal phase of the negative pass monotonically decreases for increasing T2, and the underlying multi-contrast image, uncorrupted by the background phase, is x’, which denotes voxel-wise complex conjugation of the multi-contrast image x. The corresponding forward models for positive and negative passes are

[0158] y+= FSx+= FSe^x, (5a) y

[0159]

[0160] _ = FSx_ = FSei(P^x\ (5b) Subscripts denote association with the positive and negative passes. x+= ei(p^x and x_ = ei(p^x* are phase-corrupted multi-contrast images from the positive and negative passes, respectively. One crucial assumption of this model is that the background phase is identical in both acquisitions. The phase-corrupted positive and negative pass images are computed using model-based image reconstruction:

[0161] x+= argmin||FSu — y+|||, (6a)

[0162] u x_ = argmin||FSv — y_|||. (6b)

[0163]

[0164] VModel-based image reconstruction with coil sensitivities enables undersampled43and arbitrary sampling schemes41with the PSGRE signal model. The background phase is explicitly calculated from the phases of x+and x_,

[0165] iei<p(.r)=(7)

[0166]

[0167] \x*Jv Jand the multi-contrast image is retrieved by averaging the phase-corrected images from the positive and negative passes,

[0168] x = | [x+e-^W+(x_e-^(r))*]. (8)

[0169]

[0170] Contrast extraction by (2) can then proceed without hindrance by the background phase. FIG. 1A illustrates the steps of converting positive and negative pass images to extracted T1w and T2w contrasts. FIGS. 11 A and 11B detail the effects of off-resonance on the PSGRE steady-state signal phase.

[0171] 5.2 Dual-echo UTE: bone-selective imaging

[0172] While the PSGRE signal mechanism encodes a T2-weighting along the signal phase of long-T2 species, this mechanism fails for short-T2 species like collagen-bound bone water. In a GRE, short-T2 magnetization (T2 < TR), is briefly coherent immediately following an RF pulse and experiences a rapid and near-complete decay before the pulse of the subsequent TR38’39’53Because little or no residual magnetization remains by the end of the TR, the steady-state signal behavior of short-T2 magnetization is independent of the prescribed RF phase schedule of any GRE sequence. This is shown by simulation data (FIG. 17A), which assumes ideal RF excitation efficiency (i.e. instantaneous application of an RF pulse)54’55In the simulation, the signal magnitude of long-T2 species (T2 > TR) exhibits a dependence on the RF phase increment, while that of short-T2 species does not. Instead, the short-T2 signal magnitude is predominantly a function of the transverse decay occurring within a single TR and the TE used to sample the rapidly decaying signal.A dual-echo UTE sequence is designed to explicitly probe the intra-TR decay of short-T2 signal. Like a GRE, this sequence also uses a regularly spaced train of RF pulses (typically with hard pulses of short duration and an RF phase increment 0 = 117°), but two data acquisitions are performed per TR: one at an ultrashort echo time (UTE; xUTE) and another at a conventional echo time (CTE; xCTE). The UTE acquisition is performed with minimal delay after the RF pulse (~30-60 ps), while the CTE acquisition uses a delay 4-1 Ox the T2 of bone water (~300-400 ps24), but substantially less than the T2 of soft tissue. The long-T2 components undergo negligible signal decay and appear with approximately the same signal magnitude in both echoes. Ignoring the effects of proton density on the signal magnitude, an echo subtraction |xUTE| - |xCTE| suppresses long-T2 signal and highlights short-T2 components (FIG.

[0173] 17B). Notably, it also reduces the dependence of the subtraction signal on the RF phase increment.

[0174] Tissue species with both a long-T2 and high proton density, like fat, also appear with high signal intensity in a standard echo subtraction. This is because a large proton density compensates for a small but nonzero decay in transverse magnetization between the ultrashort and conventional TEs. Since bone has low proton density (~20% by volume), the presence of a long-T2, high proton density species reduces bone specificity53. A normalized echo subtraction removes proton density weighting54and computes a bone-selective image xs.

[0175] IXUTEI—IXCTE I

[0176]

[0177] IXUTEI + lxCTE I This operation further enhances short-T2 species and minimizes the dependency of the bone-selective image signal on the RF phase increment (FIG. 17C)

[0178] Alternatively, to avoid enhancement of noise from air, which results from the nominal division by zero, a weighted least-squares method34is used to compute the normalized echo subtraction:

[0179] i i 2 xs= argmin ||W / 2(IXUTEI + lxCTE |)U - W / 2(IXUTEI IXCTE ) || ■ (10) u

[0180]

[0181] 2W is a set of weights from convolving a low-pass filter (e.g. a Hanning window) with the standard echo subtraction |xUTE| - |xCTE|. (10) is evaluated by an iterative conjugate gradient method56. Unlike (9), method (10) uses weights IV to localize voxels containing actual signal from tissue. When used in an iterative method like the conjugate gradient, these weights manipulate the convergence behavior57of the algorithm to emphasize the emergence of high intensity signal from bone and discourage signal from voxels containing air. Early termination of the conjugate gradient method yields images which are highly specific to bone while retaining the long-T2 signal suppression properties of a normalized echo subtraction34’58

[0182] Finally, because xsis derived from both xUTEand xCTE, it is beneficial to perform joint reconstruction and regularization of the UTE and CTE datasets:

[0183] (xUTE,xCTE) = argmin||FSu — yUTE||| + ||FSv - yCTE||l + R(u, v). (11)

[0184]

[0185] (w,v) A joint-L0 wavelet regularization term enhances thin bony structures because it ensures that both echoes share the same set of nonzero wavelet coefficients34. It also leverages properties of sparsity-promoting reconstruction to reduce total scan duration42 59. Typical T1w UTE and CTE images and the resulting bone-selective image are shown in FIG. 1B.

[0186] 5.3 DREAMER: Pulse Sequence and Image Reconstruction DREAMER is a pulse sequence based on the GRE framework. It uses contrast mechanisms from parent PSGRE and dual-echo UTE sequences to achieve simultaneous bone-selective, T1w, and T2w imaging. The pulse sequence diagram for DREAMER is shown in FIG. 2A. In each pass (positive and negative) a train of RF pulses is applied with a small RF phase increment to concurrently encode T1- and T2-weighting along the steady-state signal phase. Ultrashort and conventional echoes are acquired following each RF pulse to also sensitize the sequence to short-T2 signal. A single DREAMER acquisition produces four phase-corrupted images (2 echoes x 2 passes; FIG.

[0187] 2B). Analogous to a standard dual-echo sequence, which produces two T1w images and one bone-selective image, DREAMER produces two multi-contrast images (FIG. 3A) and one bone-selective image. Image contrast reconstruction from DREAMER data follows a similar path to that of the parent sequences (FIG. 3B).

[0188] DREAMER image reconstruction steps are laid out for completeness. In DREAMER image reconstruction, four phase-corrupted multi-contrast images are reconstructed from the four acquired k-space datasets:

[0189] (xk) = argmin ||FSUM- + R(uk,l). (12) fc

[0190]

[0191] ={UTE, CTE} / ={+,-} Indices fc = {UTE, CTE] and / = {+, -} are used for compactness and to denote association with the echo (ultrashort or conventional) and steady-state pass (positive or negative). The optimization (12) reconstructs a pair of positive and negative images for each of the two acquired echoes by enforcing data consistency on each k-space dataset. Like with dual-echo UTE reconstruction, a joint-L0wavelet regularization term R(uk,l) is used across all four images to promote bone specificity in the normalized echo subtraction.

[0192] Following reconstruction of the four phase-corrupted images, the background phases <puTE(r) and < PcTE(r)areestimated for both the ultrashort and conventional echoes:

[0193] i

[0194] ei< Pfc(r) = fe±¥. (13)

[0195] \xk,-J

[0196] The UTE and CTE multi-contrast images xUTEand xCTEare then similarly calculated by averaging the phase-corrected images from positive and negative passes:

[0197] = 1 [xt+e-,'p‘(r)+ (Xt-e-'w'’-’)*]. (14)

[0198]

[0199] Contrast extraction may be performed on either the UTE or CTE multi-contrast image to yield Tiw and T2W images,

[0200] X

[0201]

[0202] W = (15) and a bone-selective image xsis computed from both the UTE and CTE multicontrast images by the direct normalized echo subtraction method54(9) or by the weighted least-squares method34(10). The present study performs contrast extraction on the CTE multi-contrast image and exclusively uses the weighted least-squares method for computing the bone-selective image.5.4 Imaging details

[0203] Four healthy adult subjects (2 female; mean / median age: 26 / 25 y.o.) were recruited to receive MR imaging for this study, which investigates the feasibility of using DREAMER as a rapid, ionizing radiation-free method to simultaneously acquire Tiw, T2W, and bone-selective images. Imaging was performed at the head, knee, calf, and foot. In addition, one pediatric patient (male, 17 y.o., with a clinical history of sustained trauma to the face three years ago), indicated for a reduced-dose head CT in preparation for a delayed septoplasty, submucosal resection of the septum, and turbinate reduction at the Children’s Hospital of Philadelphia, was recruited for additional research MRI to assess the solid- and soft-tissue contrast of DREAMER. CT images were acquired with an in-plane voxel size of 0.42 mm2and 0.75 mm slice thickness. All research imaging was performed after obtaining informed consent and in accordance with the University of Pennsylvania and Children’s Hospital of Philadelphia’s Institutional Review Board requirements.

[0204] MRI was performed at 3 Tesla (Prisma, Siemens, Erlangen, Germany). DREAMER was implemented in SequenceTree60with a golden-angle ordering61and a center-out, ramp-up radial trajectory. To assess the soft-tissue contrast of DREAMER with clinical imaging methods, Tiw MPRAGE (“magnetization-prepared rapid gradient echo”) and Tiw and T2W FSE (“fast spin-echo”) were compared to their counterparts in DREAMER. The acquisition voxel sizes were matched to that of DREAMER. For the contrast comparisons, phase modulation values were selected to create a qualitative “best match” with the clinical standard. For bone-selective contrast comparison, the patient’s CT image was downsampled to an isotropic voxel size of 1 mm3and registered to match the DREAMER images. Sequence parameters for the head and for musculoskeletal anatomy are listed in Tables 1 and 2, respectively.

[0205] Image reconstruction was implemented using the SigPy62library on a desktop workstation with an Intel Core i9-10980XE, 256 GB RAM, and an NVIDIA Geforce RTX 2070 Super with 8 GB VRAM. The Flatiron Institute NUFFT was used for its computational efficiency and relatively low memoryoverhead63. Coil sensitivity maps were estimated by NLINV on the graphics processing unit43’62To accelerate convergence and reduce reconstruction time, k-space preconditioning64was used with the primal-dual hybrid gradient algorithm65, which ran on the central processing unit.

[0206] Exemplary System

[0207] FIG. 18 is a block diagram illustrating an exemplary system for controlling an MRI scanner. In Figure 18, the system includes a computing platform 1800 including at least one processor 1802 and a memory 1804. The system further includes an MR imaging controller 1806 implemented by processor 1800 for controlling an MRI scanner 1808 to perform the following steps:

[0208] (a) during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject;

[0209] (b) receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject;

[0210] (c) receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject;

[0211] (d) generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals;

[0212] (e) generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times;

[0213] (f) repeating steps (a)-(e) for a plurality of repetition times that form the MR imaging scan; and

[0214] (g) simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solidtissue images, and / or T1 -weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1- weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and solid tissues.

[0215] MR imaging controller 1806 may be implemented using computer executable instructions stored in memory 1804 and executed by processor 1802.TABLES

[0216] Table 1: Sequence parameters used for imaging at the head in this study. All sequences acquire at the same voxel size (1 mm3isotropic). Except for the voxel size and field-of-view, MPRAGE and T2W FSE sequence parameters were taken from a typical protocol used for pediatric head imaging at the Children’s Hospital of Philadelphia. All head images were acquired using a 64- channel head-and-neck coil.

[0217] Parameters DREAMER Tiw MPRAGE T2w FSE

[0218] (Healthy Subject)

[0219] FOV (280, 280, 280) mm (256, 256, 256) mm (256, 256, 75) mm

[0220] Flip angle 24° 9° 90°

[0221] RF phase increment 1.5°

[0222] Refocusing flip angle 150°

[0223] TR 5.5 ms 1900 ms 5500 ms

[0224] TE (80 ps, 2.38 ms) 2.28 ms 90 ms

[0225] Tl 958 ms

[0226] Pulse duration(s) 60 ps

[0227] Readout ramp-up time 160 ps

[0228] Dwell time OR 4 ps 190 Hz / px 260 Hz / px

[0229] BW / pixel

[0230] Matrix size (280, 280, 280) (256, 256, 256) (256, 256, 75)

[0231] Voxel size 1 mm31 mm31 mm3

[0232] GRAPPA Factor 2, w / 24 ref. lines 2, w / 24 ref. lines

[0233] Echo train le

[0234]

[0235] ngth 18Concatenations 3

[0236] Scan duration 6.0 min. (unless 4.4 min. 2.5 min.

[0237] i otherwise specified)

[0238] Note: All sequences for head imaging were acquired at the same voxel size (1 mm3isotropic). Except for the voxel size and field-of-view, MPRAGE and T2 w FSE sequence parameters were taken from a typical protocol used for pediatric head imaging at the Children's Hospital of Philadelphia. All head images were acquired using a 64-channel head-and-neck coil.

[0239] Abbreviations: FOV = “field-of-view”, Tl = “inversion time”, BW = “bandwidth”, GRAPPA = “GeneRalized Autocalibrating Partial Parallel Acquisition”, D-RF = “dual radiofrequency”, VS = “view-sharing”, px = “pixel”, ref. = “reference”, w / = “with”Table 2: Sequence parameters used for imaging musculoskeletal anatomy (knee, calf, and foot) in this study. All sequences acquire at approximately the same voxel size (0.75 mm3isotropic). Except for the voxel size, field-of-view, and scan duration, Tiw and T2W FSE parameters were taken from a typical protocol used for imaging the knee and ankle at the Hospital of the University of Pennsylvania. Knee and calf images were acquired with a 15-channel knee coil, and foot images were acquired with a 20- channel head-and-neck coil.

[0240] Parameters DREAMER Tiw FSE T2W FSE Tiw FSE T2W FSE

[0241] Healthy Subject) (Knee) (Knee) (Foot) (Foot)

[0242] FOV (279, 279, "279) (192, 192? (192, 192, (288, 288, (288, "288," mm 102.4) mm 102.4) mm 102.4) mm 102.4) mm

[0243] Flip angle 16° 90° 90° 90° 90°

[0244] RF phase incr

[0245]

[0246] ement 1.3°

[0247] Refocusing flip angle 150° 150° 127° 135°

[0248] TR i 5.35 ms 710 ms 3950 ms 710 ms 3950 ms TE (70 ps, 2.38 ms) 12 ms 74 ms 12 ms 74 ms

[0249] Pulse duration(s) 40 ps

[0250] Readout ramp-up | 160 ps

[0251] time

[0252] Dwell time OR 4 ps 250 Hz / px 255 Hz / px 250 Hz / px 255 Hz / Px BW / pixel

[0253] Matrix size (372, 372, 372) (256, 256, 128) (256, 256, 128) (384, 384, 128) (384, 384, 128) Voxel size 0.75 mm3(0.75, 0.75, 0.8) (0.75, 0.75, 0.8) (0.75, 0.75, 0.8) (0.75, 0.75, 0.8)

[0254] mm mm mm mm

[0255] GRAPPA Factor 2, w / 24 ref. 2, w / 24 ref. 2, w / 24 ref. 2, w / 24 ref.

[0256] lines lines lines linesEcho train length 3 18 3 18 Concatenations 9 8 9 8

[0257] Scan duration | 8.1 min. 5.1 min. 4.8 min. 7.4 min. 6.9 min.

[0258] Abbreviations: FOV = “field-of-view”, Tl = “inversion time”, GRAPPA = “GeneRalized Autocalibrating Partial Parallel Acquisition”, BW = “bandwidth”, D-RF = “dual radiofrequency”, VS = “view-sharing”, px = “pixel”, ref. = “reference”, w / = “with”References

[0259] The disclosure of each of the references listed below is hereby incorporated herein by reference in its entirety.

[0260] 1. Papanicolas I, Woskie LR, Jha AK. Health Care Spending in the United States and Other High-Income Countries. JAMA. 2018;319(10): 1024-39. doi: 10.1001 / jama.2018.1150. PubMed PMID: 29536101.

[0261] 2. Rohman M. JAMA: U. S. spends the most on healthcare — and imaging is a reason why: Health Imaging; 2018 [cited 2024 October 11 ]. Available from: https: / / healthimaging.com / topics / practice-management / jama-us-spends-most-healthcare-and-imaging-reason-why.

[0262] 3. Smith-Bindman R, Kwan ML, Marlow EC, Theis MK, Bolch W, Cheng SY, Bowles EJA, Duncan JR, Greenlee RT, Kushi LH, Pole JD, Rahm AK, Stout NK, Weinmann S, Miglioretti DL. Trends in Use of Medical Imaging in US Health Care Systems and in Ontario, Canada, 2000-2016. JAMA.

[0263] 2019;322(9):843-56. doi: 10.1001 / jama.2019.11456. PubMed PMID: 31479136; PMCID: PMC6724186.

[0264] 4. Fernandez E. Medical Imaging Rates Continue to Rise Despite Push to Reduce Their Use: University of California San Francisco; 2019 [cited 2024 October 11], Available from: https: / / www.ucsf.edu / news / 2019 / 09 / 415286 / medical-imaging-rates-continue-rise-despite-push-reduce-their-use.

[0265] 5. Rubin GD. Computed tomography: revolutionizing the practice of medicine for 40 years. Radiology. 2014;273(2 Suppl): S45-74. doi: 10.1148 / radiol.14141356. PubMed PMID: 25340438.

[0266] 6. Bitar R, Leung G, Perng R, Tadros S, Moody AR, Sarrazin J, McGregor C, Christakis M, Symons S, Nelson A, Roberts TP. MR pulse sequences: what every radiologist wants to know but is afraid to ask. Radiographics.

[0267] 2006;26(2):513-37. doi: 10.1148 / rg.262055063. PubMed PMID: 16549614.

[0268] 7. Falke TH. Essentials of clinical MRI: Springer Science & Business Media; 2012.

[0269] 8. Preston DC. Magnetic Resonance Imaging (MRI) of the Brain and Spine: Basics: Case Western Reserve University Department of Neurology;2006 [cited 2024 October 12], Available from: https: / / case.edu / med / neurology / NR / MRI%20Basics.htm.

[0270] 9. Zhang R, Narra VR, Kansagra AP. Large-Scale Assessment of Scan-Time Variability and Multiple-Procedure Efficiency for Cross-Sectional Neuroradiological Exams in Clinical Practice. J Digit Imaging. 2020;33(1): 143-50. doi: 10.1007 / s10278-019-00252-w. PubMed PMID: 31292770; PMCID: PMC7064664.

[0271] 10. Pearce MS, Salotti JA, Little MP, McHugh K, Lee C, Kim KP, Howe NL, Ronckers CM, Rajaraman P, Sir Craft AW, Parker L, Berrington de Gonzalez A. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study. Lancet.

[0272] 2012;380(9840):499-505. Epub 20120607. doi: 10.1016 / s0140-6736(12)60815-0. PubMed PMID: 22681860; PMCID: PMC3418594.

[0273] 11. Badve CA, K MM, Iyer RS, Ishak GE, Khanna PC. Craniosynostosis: imaging review and primer on computed tomography. Pediatr Radiol.

[0274] 2013;43(6):728-42; quiz 5-7. Epub 20130502. doi: 10.1007 / s00247-013-2673-6. PubMed PMID: 23636536.

[0275] 12. Chen J, Kanekar S. Imaging of Congenital Craniofacial Anomalies and Syndromes. Clinics in Perinatology. 2022;49(3):771-90. doi: 10.1016 / j.clp.2022.04.005.

[0276] 13. Lisle D, Monsour P, Maskiell C. Imaging of craniofacial fibrous dysplasia. Journal of Medical Imaging and Radiation Oncology.

[0277] 2008;52(4):325-32. doi: https: / / doi.org / 10.1111 / j.1440-1673.2008.01963.x. 14. O'Sullivan GJ, Carty FL, Cronin CG. Imaging of bone metastasis: An update. World J Radiol. 2015;7(8):202-11. doi: 10.4329 / wjr.v7.i8.202. PubMed PMID: 26339464; PMCID: PMC4553252.

[0278] 15. Hedlund GL, Frasier LD. Neuroimaging of abusive head trauma. Forensic Science, Medicine, and Pathology. 2009;5(4):280-90. doi: 10.1007 / s12024-009-9132-6.

[0279] 16. Vazquez E, Delgado I, Sanchez-Montanez A, Fabrega A, Cano P, Martin N. Imaging abusive head trauma: why use both computed tomography and magnetic resonance imaging? Pediatric Radiology. 2014;44(4):589-603. doi: 10.1007 / s00247-014-3216-5.17. Moran DS, Evans RK, Hadad E. Imaging of lower extremity stress fracture injuries. Sports Med. 2008;38(4):345-56. doi: 10.2165 / 00007256-200838040-00005. PubMed PMID: 18348592.

[0280] 18. Provost C, Soudant M, Legrand L, Ben Hassen W, Xie Y, Soize S, Bourcier R, Benzakoun J, Edjlali M, Boulouis G. Magnetic resonance imaging or computed tomography before treatment in acute ischemic stroke: effect on workflow and functional outcome. Stroke. 2019;50(3):659-64.

[0281] 19. Reyes-Santias F, Garcia-Garcia C, Aibar-Guzman B, Garcia-Campos A, Cordova-Arevalo O, Mendoza-Pintos M, Cinza-Sanjurjo S, Portela-Romero M, Mazon-Ramos P, Gonzalez-Juanatey JR, editors. Cost Analysis of Magnetic Resonance Imaging and Computed Tomography in Cardiology: A Case Study of a University Hospital Complex in the Euro Region. Healthcare; 2023: MDPI.

[0282] 20. Hallpike L, Hawkes D. Medical image registration: an overview. Imaging. 2002;14(6):455-63.

[0283] 21. Bravin A, Coan P, Suortti P. X-ray phase-contrast imaging: from pre-clinical applications towards clinics. Phys Med Biol. 2013;58(1): R1-35. Epub 20121210. doi: 10.1088 / 0031 -9155 / 58 / 1 / R1. PubMed PMID: 23220766. 22. Pourmorteza A, Symons R, Sandfort V, Mallek M, Fuld MK, Henderson G, Jones EC, Malayeri AA, Folio LR, Bluemke DA. Abdominal Imaging with Contrast-enhanced Photon-counting CT: First Human Experience. Radiology.

[0284] 2016;279(1):239-45. Epub 20160203. doi: 10.1148 / radiol.2016152601. PubMed PMID: 26840654; PMCID: PMC4820083.

[0285] 23. Horch RA, Nyman JS, Gochberg DF, Dortch RD, Does MD. Characterization of 1H NMR signal in human cortical bone for magnetic resonance imaging. Magn Reson Med. 2010;64(3):680-7. doi: 10.1002 / mrm.22459. PubMed PMID: 20806375; PMCID: PMC2933073. 24. Seifert AC, Wehrli SL, Wehrli FW. Bi-component T2 * analysis of bound and pore bone water fractions fails at high field strengths. NMR Biomed.

[0286] 2015;28(7):861-72. Epub 20150518. doi: 10.1002 / nbm.3305. PubMed PMID: 25981785; PMCID: PMC4478152.

[0287] 25. Kamona N, Jones BC, Lee H, Song HK, Rajapakse CS, Wagner CS, Bartlett SP, Wehrli FW Cranial bone imaging using ultrashort echo-time bone-selective MRI as an alternative to gradient-echo based "black-bone" techniques. Magma. 2024;37(1):83-92. Epub 20231107. doi: 10.1007 / s10334-023-01125-8. PubMed PMID: 37934295; PMCID: PMC10923077.

[0288] 26. Eley KA, McIntyre AG, Watt-Smith SR, Golding SJ. " Black bone" MRI: a partial flip angle technique for radiation reduction in craniofacial imaging. Br J Radiol. 2012;85(1011):272-8. doi: 10.1259 / bjr / 95110289. PubMed PMID: 22391497.

[0289] 27. Johansson A, Garpebring A, Asklund T, Nyholm T. CT substitutes derived from MR images reconstructed with parallel imaging. Med Phys.

[0290] 2014;41(8):082302. doi: 10.1118 / 1.4886766. PubMed PMID: 25086551. 28. Chen Y, Ying C, Binkley MM, Juttukonda MR, Flores S, Laforest R, Benzinger TLS, An H. Deep learning-based T1 -enhanced selection of linear attenuation coefficients (DL-TESLA) for PET / MR attenuation correction in dementia neuroimaging. Magn Reson Med. 2021;86(1):499-513. Epub 20210208. doi: 10.1002 / mrm.28689. PubMed PMID: 33559218; PMCID: PMC8091494.

[0291] 29. Eshraghi Boroojeni P, Chen Y, Commean PK, Eldeniz C, Skolnick GB, Merrill C, Patel KB, An H. Deep-learning synthesized pseudo-CT for MR high-resolution pediatric cranial bone imaging (MR-HiPCB). Magn Reson Med.

[0292] 2022;88(5):2285-97. Epub 20220617. doi: 10.1002 / mrm.29356. PubMed PMID: 35713359; PMCID: PMC9420780.

[0293] 30. Linkugel AD, Wang T, Boroojeni PE, Eldeniz C, Chen Y, Skolnick GB, Commean PK, Merrill CM, Strahle JM, Goyal MS. MR Cranial Bone Imaging: Evaluation of Both Motion-Corrected and Automated Deep Learning Pseudo-CT Estimated MR Images. American Journal of Neuroradiology. 2024.

[0294] 31. Ren J, Lakshminarayanan B. Google Research. 2019. [cited 2024], Available from: https: / / research.google / blog / improving-out-of-distribution-detection-in-machine-learning-models / .

[0295] 32. Heckel R, Jacob M, Chaudhari A, Perlman O, Shimron E. Deep learning for accelerated and robust MRI reconstruction. MAGMA.

[0296] 2024;37(3):335-68. Epub 20240723. doi: 10.1007 / s10334-024-01173-8. PubMed PMID: 39042206; PMCID: PMC11316714.33. Zur Y, Wood ML, Neuringer LJ. Spoiling of transverse magnetization in steady-state sequences. Magn Reson Med. 1991;21(2):251-63. doi: 10.1002 / mrm.1910210210. PubMed PMID: 1745124.

[0297] 34. Vu BD, Kamona N, Kim Y, Ng JJ, Jones BC, Wehrli FW, Song HK, Bartlett SP, Lee H, Rajapakse CS. Three contrasts in 3 min: Rapid, high-resolution, and bone-selective UTE MRI for craniofacial imaging with automated deep-learning skull segmentation. Magn Reson Med. 2024. Epub 20240901. doi: 10.1002 / m rm.30275. PubMed PMID: 39219299.

[0298] 35. Lee H, Zhao X, Song HK, Zhang R, Bartlett SP, Wehrli FW Rapid dual-RF, dual-echo, 3D ultrashort echo time craniofacial imaging: A feasibility study. Magn Reson Med. 2019;81(5):3007-16. Epub 2018 / 12 / 20. doi: 10.1002 / mrm.27625. PubMed PMID: 30565286; PMCID: PMC6414254. 36. Wang X, Hernando D, Reeder SB. Phase-based T(2) mapping with gradient echo imaging. Magn Reson Med. 2020;84(2):609-19. Epub 20191224. doi: 10.1002 / mrm.28138. PubMed PMID: 31872470; PMCID: PMC7180093.

[0299] 37. Tamada D, Field AS, Reeder SB. Simultaneous T(1 ) -weighted and T(2) -weighted 3D MRI using RF phase-modulated gradient echo imaging. Magn Reson Med. 2022;87(4): 1758-70. Epub 20211109. doi: 10.1002 / mrm.29077. PubMed PMID: 34752639; PMCID: PMC9006307.

[0300] 38. Rahmer J, Blume U, Börnert P. Selective 3D ultrashort TE imaging: comparison of “dual-echo” acquisition and magnetization preparation for improving short-T 2 contrast. Magnetic Resonance Materials in Physics, Biology and Medicine. 2007;20:83-92.

[0301] 39. Robson MD, Gatehouse PD, Bydder M, Bydder GM. Magnetic Resonance: An Introduction to Ultrashort TE (UTE) Imaging. Journal of Computer Assisted Tomography. 2003;27(6):825-46. PubMed PMID: 00004728-200311000-00001.

[0302] 40. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magn Reson Med. 1990; 16(2): 192-225. doi: 10.1002 / mrm.1910160203. PubMed PMID: 2266841.41. Pruessmann KP, Weiger M, Bornert P, Boesiger P. Advances in sensitivity encoding with arbitrary k-space trajectories. Magn Reson Med.

[0303] 2001;46(4):638-51. doi: 10.1002 / mrm.1241. PubMed PMID: 11590639.

[0304] 42. Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magn Reson Med.

[0305] 2007;58(6): 1182-95. doi: 10.1002 / mrm.21391. PubMed PMID: 17969013. 43. Uecker M, Hohage T, Block KT, Frahm J. Image reconstruction by regularized nonlinear inversion — joint estimation ofcoil sensitivities and image content. Magnetic Resonance in Medicine. 2008;60(3):674-82.

[0306] 44. Brown RW, Cheng Y-CN, Haacke EM, Thompson MR, Venkatesan R. Magnetic resonance imaging: physical principles and sequence design: John Wiley & Sons; 2014.

[0307] 45. Wajer FTAW, Pruessmann KP, editors. Major speedup of reconstruction for sensitivity encoding with arbitrary trajectories. Proceedings of the International Society of Magnetic Resonance in Medicine; 2001; Toronto, Canada: ISMRM.

[0308] 46. Iyer SS, Ong F, Cao X, Liao C, Daniel L, Tamir JI, Setsompop K. Polynomial preconditioners for regularized linear inverse problems. SIAM Journal on Imaging Sciences. 2024; 17(1 ): 116-46.

[0309] 47. Duyn JH. Steady state effects in fast gradient echo magnetic resonance imaging. Magn Reson Med. 1997;37(4):559-68. doi: 10.1002 / mrm.1910370414. PubMed PMID: 9094078.

[0310] 48. Haskell MW, Nielsen JF, Noll DC. Off-resonance artifact correction for MRI: A review. NMR Biomed. 2023;36(5):e4867. Epub 20221214. doi: 10.1002 / nbm.4867. PubMed PMID: 36326709; PMCID: PMC10284460. 49. Pipe JG, Zwart NR, Aboussouan EA, Robison RK, Devaraj A, Johnson KO. A new design and rationale for 3D orthogonally oversampled k-space trajectories. Magn Reson Med. 2011;66(5): 1303-11. Epub 20110405. doi: 10.1002 / mrm.22918. PubMed PMID: 21469190.

[0311] 50. Robison RK, Anderson AG, 3rd, Pipe JG. Three-dimensional ultrashort echo-time imaging using a FLORET trajectory. Magn Reson Med.

[0312] 2017;78(3): 1038-49. Epub 20161024. doi: 10.1002 / mrm.26500. PubMed PMID: 27775843.51. Bernstein M. Handbook of MRI pulse sequences: Elsevier; 2004.

[0313] 52. Lamy J, de Sousa PL, editors. Sycomore-an MRI simulation toolkit. ISMRM 28th Annual Meeting, 8-13 octobre 2020, Paris; 2020.

[0314] 53. Du J, Bydder M, Takahashi AM, Carl M, Chung CB, Bydder GM. Short T2 contrast with three-dimensional ultrashort echo time imaging. Magnetic resonance imaging. 2011;29(4):470-82.

[0315] 54. Johnson EM, Vyas U, Ghanouni P, Pauly KB, Pauly JM. Improved cortical bone specificity in UTE MR Imaging2017;77(2):684-95. doi: https: / / doi. Org / 10.1002 / mrm.26160.

[0316] 55. Carl M, Bydder M, Du J, Takahashi A, Han E. Optimization of RF excitation to maximize signal and T2 contrast of tissues with rapid transverse relaxation. Magnetic resonance in medicine. 2010;64(2):481-90.

[0317] 56. Hestenes MR, Stiefel E. Methods of Conjugate Gradients for Solving. Journal of research of the National Bureau of Standards. 1952;49(6):409. 57. Shewchuk JR. An introduction to the conjugate gradient method without the agonizing pain1994.

[0318] 58. Kamona N, Ng J J, Kim Y, BT DV, Vossough A, Wagner CS, Cordray H, Lee H, Villavisanis DF, Rajapakse CS, Bartlett SP, Wehrli FW. Craniofacial Imaging of Pediatric Patients by Ultrashort Echo-Time Bone-Selective MRI in Comparison to CT. Acad Radiol. 2024. Epub 20240905. doi: 10.1016 / j.acra.2024.08.053. PubMed PMID: 39242296.

[0319] 59. Trzasko J, Manduca A. Highly undersampled magnetic resonance image reconstruction via homotopic l(0) -minimization. IEEE Trans Med Imaging. 2009;28(1): 106-21. doi: 10.1109 / TMI.2008.927346. PubMed PMID: 19116193.

[0320] 60. Magland JF, Li C, Langham MC, Wehrli FW. Pulse sequence programming in a dynamic visual environment: SequenceTree. Magn Reson Med. 2016;75(1):257-65. Epub 20150307. doi: 10.1002 / mrm.25640. PubMed PMID: 25754837; PMCID: PMC4561593.

[0321] 61. Chan RW, Ramsay EA, Cunningham CH, Plewes DB. Temporal stability of adaptive 3D radial MRI using multidimensional golden means. Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine. 2009;61(2):354-63.62. Ong F, Lustig M. SigPy: A Python Package for High Performance Iterative Reconstruction. ISMRM 27th Annual Meeting & Exhibition; Montreal, QC, Canada2019.

[0322] 63. Barnett AH, Magland J, af Klinteberg L. A parallel nonuniform fast Fourier transform library based on an “exponential of semicircle" kernel. SIAM Journal on Scientific Computing. 2019;41(5): C479-C504.

[0323] 64. Ong F, Uecker M, Lustig M. Accelerating non-Cartesian MRI reconstruction convergence using k-space preconditioning. IEEE transactions on medical imaging. 2019;39(5): 1646-54.

[0324] 65. Chambolle A, Pock T. A first-order primal-dual algorithm for convex problems with applications to imaging. Journal of mathematical imaging and vision. 2011;40:120-45.

[0325] References for FIGS. 10A to 10F, 11 A, and 11B:

[0326] 1. Gold GE, Han E, Stainsby J, Wright G, Brittain J, Beaulieu C. Musculoskeletal MRI at 3.0 T: relaxation times and image contrast. AJR Am J Roentgenol. 2004;183(2):343-51. doi: 10.2214 / ajr.183.2.1830343. PubMed PMID: 15269023.

[0327] 2. Stanisz GJ, Odrobina EE, Pun J, Escaravage M, Graham SJ, Bronskill MJ, Henkelman RM. T1, T2 relaxation and magnetization transfer in tissue at 3T. Magn Reson Med. 2005;54(3):507-12. doi: 10.1002 / mrm.20605. PubMed PMID: 16086319.

[0328] It will be understood that various details of the subject matter described herein may be changed without departing from the scope of the subject matter described herein. Furthermore, the foregoing description is for the purpose of illustration only, and not for the purpose of limitation, as the subject matter described herein is defined by the claims as set forth hereinafter.

Claims

1. CLAIMSWhat is claimed is:

1. A method for magnetic resonance (MR) imaging, the method comprising:(h) during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject;(i) receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject;(j) receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject;(k) generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals;(l) generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times;(m) repeating steps (a)-(e) for a plurality of repetition times that form the MR imaging scan; and(n) simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1 -weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1- weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subjectwith different contrast levels, and / or short T2-selective image of the MR imaging subject of soft and / or solid tissues.

2. The method of claim 1, wherein the RF pulses follow an RF phase schedule < >(n) determined by the RF phase increment 0, where n is an index corresponding to the repetition time and 0 is greater than 0 degrees but less than or equal to about 5 degrees.

3. The method of claim 2, wherein the RF phase schedule follows the formula (n)= (n-1)+n0, where < >(n) is the phase of the RF pulse for the nth repetition time and (p(n - 1) is the phase of the RF pulse for the (n-1)th repetition time.

4. The method of any one of claim 1-3, wherein receiving the first echo signal includes receiving the first echo signal at a time within a range of about 10 μs to about 100 μs after transmission of the RF pulse and receiving the second echo signal includes receiving the second echo signal at a time within a range of about 1 ms to about 4 ms after transmission of the RF pulse.

5. The method of any one of claims 1-4, wherein simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images comprises applying phase modulation, optionally according to a phase modulation angle i according to the following equation: x^ = |$R(xel^)|,whereindenotes the extracted contrast, which was encoded along an angle i measured from the real axis. 5R(u) denotes the voxel-wise real value of an arbitrary complex-valued image u, and |u| is the voxelwise absolute value of u.

6. The method of any one of claims 1-5, comprising estimating a background phase with two acquisitions, one with a positive RF phase increment and one with equal but negative RF phase increment.

7. The method of claim 6, comprising correcting images of the MR with the background phase.

8. The method of any one of claims 1 -7, wherein the T 1 -weighted images of the MR imaging subject pertain to soft tissue; T2-weighted imagesof the MR imaging subject pertain to soft tissue and / or bone tissue; and a short T2-selective image of the MR imaging subject pertains to bone tissue.

9. The method of any one of claims 1 -8, wherein the MR imaging subject is a pediatric subject.

10. The method of any one of claims 1 -9, wherein the MR images comprise accelerated, high-resolution, and / or self-registered MR images.

11. A system for controlling a magnetic resonance imaging (MRI) scanner, the system comprising:at least one processor and a memory; andan MR imaging controller implemented by the at least one processor for controlling an MR scanner to perform the following steps:(a) during a repetition time in an MR imaging scan, transmit a radio frequency (RF) pulse with a phase into an MR imaging subject; (b) receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject;(c) receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject;(d) generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals;(e) generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imaging subject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times;(f) repeating steps (a)-(e) for a plurality of repetition times that form the MR imaging scan; and(g) simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1 -weighted images of the MR imagingsubject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1- weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and solid tissues.

12. The system of claim 11, wherein the RF pulses follow an RF phase schedule < >(n) determined by the RF phase increment 0, where n is an index corresponding to the repetition time and 0 is greater than 0 degrees but less than or equal to about 5 degrees.

13. The system of any one of claims 11-12, wherein the RF phase schedule follows the formula (n)= (n-1)+n0, where < >(n) is the phase of the RF pulse for the nth repetition time and (p(n - 1) is the phase of the RF pulse for the (n-1)th repetition time.

14. The system of any one of claims 11-13, wherein receiving the first echo signal includes receiving the first echo signal at a time within a range of about 10 μs to about 100 μs after transmission of the RF pulse and receiving the second echo signal includes receiving the second echo signal at a time within a range of about 1 ms to about 4 ms after transmission of the RF pulse.

15. The system of any one of claims 11-14, wherein simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images comprises applying phase modulation, optionally according to a phase modulation angle i according to the following equation:whereindenotes the extracted contrast, which was encoded along an angle i measured from the real axis. 5R(u) denotes the voxel-wise real value of an arbitrary complex-valued image u, and |u| is the voxelwise absolute value of u..

16. The system of any one of claims 11-15, comprising estimating a background phase with two acquisitions, one with a positive RF phase increment and one with equal but negative RF phase increment.

17. The system of claim 16, comprising correcting images of the MR with the background phase.

18. The system of any one of claims 11-17, wherein the T1 -weighted images of the MR imaging subject pertain to soft tissue; T2-weighted images of the MR imaging subject pertain to soft tissue and / or bone tissue; and a short T2-selective image of the MR imaging subject pertains to bone tissue.

19. The system of any one of claims 11-18, wherein the MR imaging subject is a pediatric subject.

20. The system of any one of claims 11-19, wherein the MR images comprise accelerated, high-resolution, and / or self-registered MR images.

21. A non-transitory computer readable medium having stored thereon executable instructions that when executed by a processor of a computer control the computer to cause a magnetic resonance imaging (MRI) scanner to perform steps comprising:(a) during a repetition time in an MR imaging scan, transmitting a radio frequency (RF) pulse with a phase into an MR imaging subject;(b) receiving, during the repetition time, a first echo signal at a first time after the transmission of the RF pulse into the MR imaging subject;(c) receiving, during the repetition time, a second echo signal at a second time after transmission of the RF pulse into the MR imaging subject;(d) generating and transmitting into the MR imaging subject a magnetic gradient for encoding signal during reception of the first and second echo signals;(e) generating and transmitting, during the repetition time, after reception of the second echo signal, and into the MR imagingsubject, a spoiler gradient for dephasing residual magnetization from the repetition time and previous repetition times;(f) repeating steps (a)-(e) for a plurality of repetition times that form the MR imaging scan; and(g) simultaneously generating, from first and second echo signals received during each repetition time of MR imaging scan, a spectrum of multi-contrast MR images, such as soft and / or solid tissue images, and / or T1 -weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject, including T1- weighted images of the MR imaging subject with different contrast levels, T2-weighted images of the MR imaging subject with different contrast levels, and / or short T2-selective images of the MR imaging subject of soft and / or solid tissues.

22. The non-transitory computer readable medium of claim 21, wherein the MR images comprise accelerated, high-resolution, and / or self-registered MR images.

Images