Systems and methods for dynamic noise cancellation for MRI

Abstract

Provided herein are systems and methods for mitigating electromagnetic interference (EMI) in magnetic resonance (MR) imaging. The method includes obtaining primary MR signal data and auxiliary EMI data. A multi-dimensional noise data representation is constructed using data collected from auxiliary EMI detectors and / or MRI receive coil elements during noise-only periods. Dimensionality reduction is applied to this representation to extract a low-rank noise subspace, effectively separating deterministic environmental interference from random detector noise. A truncated basis set is generated and dynamically transformed to fit the primary MR signal data for each individual repetition time (TR) interval. A noise estimate is calculated based on this dynamic transformation and subtracted from the primary MR signal data to generate EMI-corrected MR signal data. This dynamic, per-TR subspace extraction avoids rigid k-space trajectory assumptions, enabling robust noise cancellation across diverse spatial encoding trajectories, pulse sequences, and hardware configurations.

Description

WSGR Docket No. 49880-719601SYSTEMS AND METHODS FOR DYNAMIC NOISE CANCELLATION FOR MRI CROSS-REFERENCE

[0001] This PCT application claims the benefit of U.S. Provisional Application No. 63 / 778,995, filed March 27, 2025, which application is incorporated herein by reference.BACKGROUND OF THE INVENTION

[0002] Magnetic resonance imaging (MRI) is highly susceptible to electromagnetic interference (EMI), which can severely degrade image quality and diagnostic accuracy. EMI introduces substantial negative effects into magnetic resonance images, including increased background noise and image artifacts such as zippers, spikes, and DC offsets. These artifacts hinder the interpretation of underlying anatomical information and may be mistaken for anatomical structures, leading to potential misdiagnoses. Interference occurs when extraneous radiofrequency (RF) signals couple into the MRI receive chain. Common EMI sources include broadcast signals, medical monitoring equipment, fluorescent lighting, power-line harmonics, wireless communication signals, and internal electronic components housed within the scanner itself, such as high-power switching transients from gradient power supplies and RF amplifiers.SUMMARY OF THE INVENTION

[0003] Provided herein is a method for mitigating electromagnetic interference (EMI) in magnetic resonance (MR) imaging. The method can comprise obtaining EMI data from at least one EMI detector. The method can comprise obtaining MR signal data associated with an object within an MR imaging region. The method can comprise building a noise convolution matrix from the obtained EMI data. The method can comprise extracting a noise subspace from the noise convolution matrix by computing a low-rank approximation, thereby generating a truncated basis set representative of dominant EMI components. The method can comprise projecting the noise convolution matrix onto the truncated basis set to generate a reduced-dimensionality noise representation. The method can comprise computing a dynamic transformation individually for each repetition time (TR) interval to map the reduced-dimensionality noise representation to the obtained MR signal data, thereby generating a noise estimate. The method can comprise subtracting the noise estimate from the obtained MR signal data to generate corrected MR signal data.

[0004] In some cases, the at least one EMI detector comprises one or more of an E-field probe, a surface coil, an electrode connected to the patient's skin, a direct connection to a sub-system ground, a direct connection to a patient table, or a 50 Ohm terminated channel. In some cases, the at least one EMI detector is positioned external to the MR imaging volume. In some cases,WSGR Docket No. 49880-719601the obtained EMI data comprises data collected using one or more primary MRI receive (Rx) coil elements during a known noise region of an MR pulse sequence.

[0005] In some cases, building the noise convolution matrix comprises constructing a multidimensional block Hankel convolution matrix using a sliding window operation. In some cases, the EMI data is obtained from an MR pulse sequence during a predetermined noise-only acquisition region. In some cases, computing the low-rank approximation comprises applying a dimensionality reduction technique comprising one or more of: singular value decomposition (SVD), principal component analysis (PCA), eigenvalue decomposition, or randomized low-rank matrix approximation.

[0006] In some cases, generating the truncated basis set comprises thresholding singular values or eigenvalues derived from the dimensionality reduction technique based on a predefined fractional threshold. In some cases, computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors and Rx coil elements utilizing data acquired from noise-only periods. In some cases, computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors utilizing only data acquired during signal portions of MRI data acquisitions. In some cases, the noise whitening procedure is applied prior to computing the low-rank approximation using singular value decomposition (SVD).

[0007] In some cases, the at least one EMI detector comprises noise coils tuned to frequencies associated with excited bands. In some cases, the noise coils are configured to have high sensitivity to selected frequencies, thereby comprising a high Q-factor. In some cases, the noise coils are tuned over a broader frequency band and comprise a sensitivity versus frequency profile that matches a frequency distribution of an EMI spectrum over the broader frequency band. In some cases, a noise correlation matrix and subsequent low-rank approximation are re-computed for specified frequency bands having bandwidths equal to a bandwidth of excited slabs, portions of an excited bandwidth corresponding to slices in a z-dimension, or portions dominated by a particular EMI noise source.

[0008] In some cases, the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only time domain data acquired in a non-excited data acquisition. In some cases, the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only portions of time domain data acquired during an MRI data acquisition. In some cases, the noise correlation matrix is determined using a sliding window FT on the noise-only time domain data. In some cases, a sliding window time length is determined by a bandwidth of dominant noise sources or a time length required to obtain statistically significant noise correlation information.WSGR Docket No. 49880-719601In some cases, wherein the sliding window FT employs zero filling on negative and positive sides of a time window to improve frequency resolution. In some cases, a noise correlation computation is applied to data acquired during signal portions of an MR pulse sequence by employing a signal subtraction method to remove MR signal from a signal echo, thereby leaving noise-only data to perform the noise correlation computation using both MRI Rx coil elements and the at least one EMI detector. In some cases, computing the dynamic transformation comprises fitting the reduced-dimensionality noise representation to the obtained MR signal data using a least-squares estimation to determine transformation coefficients for each individual TR interval.

[0009] The method can further comprise reconstructing an EMI-corrected MR image using the corrected MR signal data. In some cases, computing the noise estimate and subtracting the noise estimate are performed jointly within an iterative image reconstruction process. In some cases, the iterative image reconstruction process comprises minimizing a joint objective function that simultaneously estimates a reconstructed MR image and the noise estimate. In some cases, the obtained MR signal data comprises k-space data acquired using either a fully sampled acquisition trajectory or an undersampled acquisition trajectory. In some cases, the joint objective function comprises: a data fidelity term that evaluates a difference between the obtained MR signal data and a predicted MR signal derived from the reconstructed MR image and the noise estimate; an image regularization term configured to enforce structural properties, on the reconstructed MR image; and a noise regularization term configured to enforce a low-rank constraint on the extracted noise subspace. In some cases, the structural properties comprise sparsity or smoothness. In some cases, the noise regularization term comprises a nuclear norm penalty applied to the noise convolution matrix. In some cases, minimizing the joint objective function comprises executing an alternating minimization algorithm that alternates between updating the reconstructed MR image and updating the noise estimate. In some cases, updating the noise estimate comprises applying singular value thresholding (SVT) to a data residual to dynamically update the truncated basis set at each iteration of the alternating minimization algorithm.

[0010] The method can further comprise performing a pre-scan calibration to identify one or more dominant noise frequency bands within the MR imaging environment; and adaptively tuning the at least one EMI detector to the identified one or more dominant noise frequency bands prior to obtaining the EMI data. In some cases, adaptively tuning the at least one EMI detector comprises applying a hardware or software-based bandpass filter configured to maximize sensitivity to the identified one or more dominant noise frequency bands.WSGR Docket No. 49880-719601

[0011] In some cases, the truncated basis set is derived at least in part from a historical noise subspace model, wherein the historical noise subspace model is constructed from accumulated EMI data obtained during a plurality of prior MR imaging sessions at a specific installation site.

[0012] In some cases, extracting the noise subspace is performed prior to obtaining the MR signal data by computing the low-rank approximation on the accumulated EMI data, thereby generating a pre-computed site-specific basis set configured to bypass real-time basis set generation during a current MR imaging session. In some cases, computing the noise estimate comprises a two-pass estimation process, comprising: a first pass that projects the noise convolution matrix onto the historical noise subspace model to remove baseline environmental interference; and a second pass that projects a residual noise representation onto a scan-specific basis set derived from the EMI data obtained during the current MR imaging session to remove transient interference.

[0013] The method can further comprise comparing a rank or a singular value distribution of the extracted noise subspace to a historical baseline noise subspace model; determining a deviation metric between the extracted noise subspace and the historical baseline noise subspace model; and triggering an automated system action if the deviation metric exceeds a predetermined threshold. In some cases, the automated system action comprises one or more of generating a user alert indicating elevated environmental interference, dynamically increasing a number of signal averages (NEX) for the MR pulse sequence, or switching to an alternative noise cancellation protocol.

[0014] In some cases, extracting the noise subspace comprises processing the noise convolution matrix through a trained artificial neural network comprising an undercomplete autoencoder architecture, wherein a bottleneck layer of the undercomplete autoencoder is configured to output a latent representation that defines the truncated basis set.

[0015] In some cases, the iterative image reconstruction process is executed by a trained unrolled neural network, wherein individual layers or blocks of the trained unrolled neural network correspond to iterations of an alternating minimization algorithm, and wherein the trained unrolled neural network comprises learnable parameters configured to dynamically apply singular value thresholding (SVT) to extract the noise subspace. In some cases, the trained unrolled neural network is trained using a supervised learning protocol comprising: a training dataset comprising pairs of EMI-corrupted MR k-space data and corresponding ground-truth unaliased MR images; and a loss function configured to minimize a difference between an output of the trained unrolled neural network and the ground-truth unaliased MR images.WSGR Docket No. 49880-719601

[0016] In some cases, the MR signal data is acquired using a low-field or ultra-low-field MR scanner operating in an unshielded environment lacking a stationary Faraday cage. In some cases, the MR signal data is acquired using a single-sided MR scanner. In some cases, the MR signal data is acquired using a portable MR scanner configured for point-of-care operation. In some cases, the MR signal data is acquired using a pulse sequence comprising non-linear spatiotemporal spatial encoding trajectories.

[0017] In some cases, the MRI signal used for subtraction is estimated using signals received by a plurality of receive coils and spatial sensitivity relationships between the receive coils. In some cases, estimating the MRI signal includes reconstructing an MRI image and using coil sensitivity profiles to estimate signal contributions in each receive channel. In some cases, the estimated signal contributions are subtracted from the received signals to generate improved estimates of electromagnetic interference present in each receive channel, and wherein the steps of signal estimation, signal subtraction, and noise estimation are repeated iteratively to refine estimates of MRI signal and electromagnetic interference. In some cases, coil sensitivity profiles are estimated or refined using the reconstructed MRI signal and residual signals obtained after signal subtraction. In some cases, spatial signatures of electromagnetic interference across the receive coils are estimated and used to suppress interference components.

[0018] In some cases, obtaining the MR signal data comprises using a magnetic resonance imaging (MRI) system comprising: a housing comprising a surface for contact with a subject; and a radio frequency receive (RF RX) coil network. In some cases, the RF RX coil network is configured to enable imaging in a region of interest. In some cases, the region of interest is external to the surface of the housing by a distance ranging from about 80 mm to about 120 mm.

[0019] In some cases, the RF RX coil network is configured to fully cover the surface such that there is no access aperture on the surface nearest the region of interest. In some cases, the RF RX coil network is configured for imaging external to the surface by a distance of about 100 mm. In some cases, the RF RX coil network comprises a plurality of RF RX coils. In some cases, the RF RX coil network comprises a plurality of interconnected RF RX coils. In some cases, the RF RX coil network comprises a plurality of coupled RF RX coils. In some cases, a number of turns and loops of the RF RX coil is configured to be adjustable to cover an entire space between legs of the subject such that the region of interest is entirely or partially covered. In some cases, the housing further comprises a radio frequency transmit (RF TX) coil proximate to the surface of the housing, wherein the RF TX coil is configured to generate an electromagnetic field in the region of interest.WSGR Docket No. 49880-719601

[0020] In some cases, the RF TX coil comprises a plurality of figure-8 coils arranged proximal to the surface. In some cases, the plurality of figure-8 coils are configured to generate a varying magnetic RF field within the region of interest. In some cases, the plurality of figure-8 coils are orthogonal to each other. In some cases, the plurality of figure-8 coils are tunable to same radiofrequency (RF) resonant frequencies. In some cases, the plurality of figure-8 coils are tunable to different RF resonant frequencies. In some cases, the plurality of figure-8 coils are configured to generate a uniform magnetic RF field within the region of interest. In some cases, the MRI system further comprises an electromagnet configured to generate an electromagnetic field in the region of interest. In some cases, the housing further comprises a gradient coil set positioned proximate to the surface, wherein the gradient coil set is configured to generate an electromagnetic field in the region of interest. In some cases, the gradient coil set comprises a single-sided gradient coil set.

[0021] In some cases, the MRI system is configured to be used for one or more of diagnosis, grading, treatment planning, or monitoring of pelvic conditions. In some cases, the MRI system comprises a magnetic field strength of less than about 0.5 T. In some cases, the MRI system comprises one or more of an open or single-sided MRI. In some cases, the housing comprises a through-bore access aperture. In some cases, the housing does not comprise a through-bore access aperture. In some cases, the MRI system is configured to be used in an office setting without shielding or floor reinforcements. In some cases, the MRI system comprises at least one permanent magnet configured for use without superconducting material.

[0022] In some cases, the RF RX coil is configured to capture images of the subject when the subject is in a position in front of or on top of the MRI, wherein the position is a high lithotomy, an inclined lithotomy, or a seated position over the MRI. In some cases, the RF RX coil is configured to capture images of the subject when the subject is in contact with the surface of the MRI in the high lithotomy, the inclined lithotomy, or the seated position. In some cases, the housing is configured to be positioned such that a central axis thereof is perpendicular to a floor when capturing images of the subject in the high lithotomy or the inclined lithotomy. In some cases, the housing is configured to be positioned such that a central axis thereof is parallel to a floor when capturing images of the subject in the seated position. In some cases, the MRI system is configured to be usable in a first mode and a second mode, wherein the first mode comprises capturing images of the subject in the high lithotomy or the inclined lithotomy when the housing is positioned such that a central axis thereof is perpendicular to a floor, and wherein the second mode comprises capturing images of the subject in the seated position when the housing is positioned such that a central axis thereof is parallel to the floor.WSGR Docket No. 49880-719601

[0023] Disclosed herein is a system for mitigating electromagnetic interference (EMI) in magnetic resonance (MR) imaging. The system can comprise at least one computer processor. The system can comprise at least one EMI detector configured to obtain EMI data and convey the obtained EMI data to the at least one computer processor. The system can comprise a magnetic resonance imaging device configured to image an object within an MR imaging region and obtain MR signal data of the object. The system can comprise one or more non-transitory computer-readable storage media storing instructions that, when executed by the at least one processor, cause the system to: build a noise convolution matrix from the obtained EMI data; extract a noise subspace from the noise convolution matrix by computing a low-rank approximation, thereby generating a truncated basis set representative of dominant EMI components; project the noise convolution matrix onto the truncated basis set to generate a reduced-dimensionality noise representation; compute a dynamic transformation individually for each repetition time (TR) interval to map the reduced-dimensionality noise representation to the obtained MR signal data, thereby generating a noise estimate; and subtract the noise estimate from the obtained MR signal data to generate corrected MR signal data.

[0024] In some cases, the at least one EMI detector comprises one or more of: an E-field probe, a surface coil, an electrode connected to the patient's skin, a direct connection to a sub-system ground, a direct connection to a patient table, or a 50 Ohm terminated channel. In some cases, the at least one EMI detector is positioned external to the MR imaging volume. In some cases, the obtained EMI data comprises data collected using one or more primary MRI receive (Rx) coil elements during a known noise region of an MR pulse sequence. In some cases, building the noise convolution matrix comprises constructing a multi-dimensional block Hankel convolution matrix using a sliding window operation. In some cases, the EMI data is obtained from an MR pulse sequence during a predetermined noise-only acquisition region.

[0025] In some cases, computing the low-rank approximation comprises applying a dimensionality reduction technique comprising one or more of: singular value decomposition (SVD), principal component analysis (PCA), eigenvalue decomposition, or randomized low-rank matrix approximation. In some cases, generating the truncated basis set comprises thresholding singular values or eigenvalues derived from the dimensionality reduction technique based on a predefined fractional threshold. In some cases, computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors and Rx coil elements utilizing data acquired from noise-only periods. In some cases, computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors utilizing only data acquired during signal portions of MRI data acquisitions. In some cases, the noise whiteningWSGR Docket No. 49880-719601procedure is applied prior to computing the low-rank approximation using singular value decomposition (SVD).

[0026] In some cases, the at least one EMI detector comprises noise coils tuned to frequencies associated with excited bands. In some cases, the noise coils are configured to have high sensitivity to selected frequencies, thereby comprising a high Q-factor. In some cases, the noise coils are tuned over a broader frequency band and comprise a sensitivity versus frequency profile that matches a frequency distribution of an EMI spectrum over the broader frequency band. In some cases, a noise correlation matrix and subsequent low-rank approximation are re-computed for specified frequency bands having bandwidths equal to a bandwidth of excited slabs, portions of an excited bandwidth corresponding to slices in a z-dimension, or portions dominated by a particular EMI noise source. In some cases, the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only time domain data acquired in a non-excited data acquisition. In some cases, the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only portions of time domain data acquired during an MRI data acquisition. In some cases, the noise correlation matrix is determined using a sliding window FT on the noise-only time domain data. In some cases, a sliding window time length is determined by a bandwidth of dominant noise sources or a time length required to obtain statistically significant noise correlation information. In some cases, the sliding window FT employs zero filling on negative and positive sides of a time window to improve frequency resolution.

[0027] In some cases, a noise correlation computation is applied to data acquired during signal portions of an MR pulse sequence by employing a signal subtraction system to remove MR signal from a signal echo, thereby leaving noise-only data to perform the noise correlation computation using both MRI Rx coil elements and the at least one EMI detector. In some cases, computing the dynamic transformation comprises fitting the reduced-dimensionality noise representation to the obtained MR signal data using a least-squares estimation to determine transformation coefficients for each individual TR interval. In some cases, the instructions are further configured to cause the system to reconstruct an EMI-corrected MR image from the corrected MR signal data. In some cases, computing the noise estimate and subtracting the noise estimate are performed jointly within an iterative image reconstruction process. In some cases, the iterative image reconstruction process comprises minimizing a joint objective function that simultaneously estimates a reconstructed MR image and the noise estimate.

[0028] In some cases, the obtained MR signal data comprises k-space data acquired using either a fully sampled acquisition trajectory or an undersampled acquisition trajectory. In some cases,WSGR Docket No. 49880-719601the joint objective function comprises: a data fidelity term that evaluates a difference between the obtained MR signal data and a predicted MR signal derived from the reconstructed MR image and the noise estimate; an image regularization term configured to enforce structural properties, on the reconstructed MR image; and a noise regularization term configured to enforce a low-rank constraint on the extracted noise subspace. In some cases, the structural properties comprise sparsity or smoothness. In some cases, the noise regularization term comprises a nuclear norm penalty applied to the noise convolution matrix.

[0029] In some cases, minimizing the joint objective function comprises executing an alternating minimization algorithm that alternates between updating the reconstructed MR image and updating the noise estimate. In some cases, updating the noise estimate comprises applying singular value thresholding (SVT) to a data residual to dynamically update the truncated basis set at each iteration of the alternating minimization algorithm. In some cases, the instructions are further configured to cause the system to: perform a pre-scan calibration to identify one or more dominant noise frequency bands within the MR imaging environment; and adaptively tune the at least one EMI detector to the identified one or more dominant noise frequency bands prior to obtaining the EMI data. In some cases, adaptively tuning the at least one EMI detector comprises applying a hardware or software-based bandpass filter configured to maximize sensitivity to the identified one or more dominant noise frequency bands.

[0030] In some cases, the truncated basis set is derived at least in part from a historical noise subspace model, wherein the historical noise subspace model is constructed from accumulated EMI data obtained during a plurality of prior MR imaging sessions at a specific installation site. In some cases, extracting the noise subspace is performed prior to obtaining the MR signal data by computing the low-rank approximation on the accumulated EMI data, thereby generating a pre-computed site-specific basis set that bypasses real-time basis set generation during a current MR imaging session.

[0031] In some cases, computing the noise estimate comprises a two-pass estimation process, comprising: a first pass that projects the noise convolution matrix onto the historical noise subspace model to remove baseline environmental interference; and a second pass that projects a residual noise representation onto a scan-specific basis set derived from the EMI data obtained during the current MR imaging session to remove transient interference.

[0032] In some cases, the instructions are further configured to cause the system to: compare a rank or a singular value distribution of the extracted noise subspace to a historical baseline noise subspace model; determine a deviation metric between the extracted noise subspace and theWSGR Docket No. 49880-719601historical baseline noise subspace model; and trigger an automated system action if the deviation metric exceeds a predetermined threshold.

[0033] In some cases, the automated system action comprises one or more of generating a user alert indicating elevated environmental interference, dynamically increasing a number of signal averages (NEX) for the MR pulse sequence, or switching to an alternative noise cancellation protocol. In some cases, extracting the noise subspace comprises processing the noise convolution matrix through a trained artificial neural network comprising an undercomplete autoencoder architecture. In some cases, a bottleneck layer of the undercomplete autoencoder is configured to output a latent representation that defines the truncated basis set.

[0034] In some cases, the iterative image reconstruction process is executed by a trained unrolled neural network. In some cases, individual layers or blocks of the trained unrolled neural network correspond to iterations of an alternating minimization algorithm. In some cases, the trained unrolled neural network comprises learnable parameters configured to dynamically apply singular value thresholding (SVT) to extract the noise subspace. In some cases, the trained unrolled neural network is trained using a supervised learning protocol comprising: a training dataset comprising pairs of EMI-corrupted MR k-space data and corresponding ground-truth unaliased MR images; and a loss function configured to minimize a difference between an output of the trained unrolled neural network and the ground-truth unaliased MR images.

[0035] In some cases, the MR signal data is acquired using a low-field or ultra-1 ow-fi eld MR scanner operating in an unshielded environment lacking a stationary Faraday cage. In some cases, the MR signal data is acquired using a single-sided MR scanner. In some cases, the MR signal data is acquired using a portable MR scanner configured for point-of-care operation. In some cases, the MR signal data is acquired using a pulse sequence comprising non-linear spatiotemporal spatial encoding trajectories. In some cases, the MRI signal used for subtraction is estimated using signals received by a plurality of receive coils and spatial sensitivity relationships between the receive coils.

[0036] In some cases, estimating the MRI signal includes reconstructing an MRI image and using coil sensitivity profiles to estimate signal contributions in each receive channel. In some cases, the estimated signal contributions are subtracted from the received signals to generate improved estimates of electromagnetic interference present in each receive channel. In some cases, the steps of signal estimation, signal subtraction, and noise estimation are repeated iteratively to refine estimates of MRI signal and electromagnetic interference. In some cases, coil sensitivity profiles are estimated or refined using the reconstructed MRI signal and residualWSGR Docket No. 49880-719601signals obtained after signal subtraction. In some cases, spatial signatures of electromagnetic interference across the receive coils are estimated and used to suppress interference components.

[0037] Another aspect of the present disclosure provides a system comprising one or more computer processors and computer memory coupled thereto. The computer memory comprises machine executable code that, upon execution by the one or more computer processors, implements any of the methods above or elsewhere herein.

[0038] Additional aspects and advantages of the present disclosure will become readily apparent to those skilled in this art from the following detailed description, wherein only illustrative embodiments of the present disclosure are shown and described. As will be realized, the present disclosure is capable of other and different embodiments, and its several details are capable of modifications in various obvious respects, all without departing from the disclosure.Accordingly, the drawings and description are to be regarded as illustrative in nature, and not as restrictive.INCORPORATION BY REFERENCE

[0039] All publications, patents, and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference.BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present invention will be obtained by reference to the following detailed description that sets forth illustrative embodiments, in which the principles of the invention are utilized, and the accompanying drawings of which:

[0041] FIG. 1 shows a schematic illustration of an example MRI system architecture configured to acquire primary MR signal data and auxiliary EMI data, in accordance with example embodiments described herein.

[0042] FIG. 2 shows a flowchart illustration of an example method for mitigating EMI in MR imaging using a pre-processing subspace extraction approach, in accordance with example embodiments described herein.

[0043] FIG. 3 shows a schematic illustration of an example method for jointly estimating a reconstructed MR image and a low-rank EMI noise representation using an iterative alternating minimization algorithm, in accordance with example embodiments described herein.

[0044] FIG. 4 is a conceptual data structure diagram illustrating the mathematical matrix transformations, including the construction of a block Hankel matrix and subsequent singularWSGR Docket No. 49880-719601value thresholding, used to extract the noise subspace, in accordance with example embodiments described herein.

[0045] FIG. 5 shows an example illustration of three EMI sensors oriented orthogonally in an auxiliary box in accordance with various embodiments described herein. FIG. 6 shows a perspective view of a non-limiting example of a female patient being imaged in a high lithotomy, in accordance with example embodiments described herein.

[0046] FIG. 7 shows a perspective view of a non-limiting example of a female patient being imaged in a seated position, in accordance with example embodiments described herein.

[0047] FIG. 8A shows a schematic illustration of a magnetic resonance imaging system, in accordance with various embodiments.

[0048] FIG. 8B illustrates an exploded view of the magnetic resonance imaging system shown in FIG. 8A in accordance with example embodiments described herein.

[0049] FIG. 8C shows a schematic front view of the magnetic resonance imaging system shown in FIG. 8A, in accordance with various embodiments.

[0050] FIG. 8D shows a schematic side view of the magnetic resonance imaging system shown in FIG. 8A, in accordance with various embodiments.

[0051] FIG. 9 shows a schematic view of an implementation of a magnetic imaging apparatus, according to various embodiments.

[0052] FIG. 10 shows a schematic view of an implementation of a magnetic imaging apparatus, according to various embodiments.

[0053] FIG. 11 shows a schematic front view of a magnetic resonance imaging system, according to various embodiments.

[0054] FIG. 12A shows an example schematic illustration of a radio frequency receive coil (RF-RX) array including individual coil elements, in accordance with various embodiments.

[0055] FIG. 12B shows an example illustration of a loop coil along with example calculations for a loop coil magnetic field, in accordance with various embodiments.

[0056] FIG. 12C shows an example X-Y chart illustrating the magnetic field as a function of radius of a loop coil, in accordance with various embodiments disclosed herein.

[0057] FIG. 12D shows a cross-sectional illustration of a portion of the human body, namely in the area of the prostate, in accordance with example embodiments described herein.

[0058] FIGS. 13A-13X illustrate various positions of patient depending on the type of anatomical scan for imaging in a magnetic resonance imaging system, according to various embodiments.WSGR Docket No. 49880-719601

[0059] FIG. 14 shows a schematic view of figure-8 coils described herein disposed within the housing of an MRI system in accordance with example embodiments described herein.

[0060] FIG. 15 shows a flowchart for a method of performing a scan on a magnetic resonance imaging system, according to various embodiments.

[0061] FIG. 16 shows a flowchart for another method of performing a scan on a magnetic resonance imaging system, according to various embodiments.

[0062] FIG. 17 shows an example computer system that is programmed or otherwise configured to implement methods provided herein in accordance with example embodiments described herein.

[0063] FIG. 18 shows exemplary NMR spectrum data for a single primary receive coil collected in phantom with and without the presence of EMI introduced using a signal generator in accordance with example embodiments described herein.DETAILED DESCRIPTION OF THE INVENTION

[0064] Stationary, high-field MRI scanners may rely on passive shielding via stationary Faraday cages to account for EMI. While effective against external EMI, Faraday cages may impose significant limitations on a system's clinical footprint, weight, and installation cost, and they provide no mitigation for internally generated EMI. For MRI systems designed to be lower cost or non-stationary, such as low-field (LF), ultra-1 ow-fi eld (ULF), and portable MRI scanners designed for point-of-care applications, less effective passive shielding techniques (e.g., applying a conductive shield close to the scanner or encompassing the patient in a mesh cloth) may be utilized. However, these techniques may limit patient access, which is crucial for interventional procedures. The signal-to-noise ratio (SNR) is inherently lower in LF and ULF systems compared to high-field counterparts, so the presence of EMI can be particularly detrimental, necessitating robust active and post-processing EMI cancellation techniques.

[0065] Several active and post-processing approaches to EMI suppression have been proposed but exhibit critical limitations. Methods utilizing Wiener filter deconvolution or adaptive filtering (e.g., least mean square or recursive least square algorithms) may often suffer from large convergence times and computational complexity. Furthermore, these methods may frequently rely on calibration data measured during dummy cycles preceding the normal scan, severely hindering their ability to mitigate dynamic, time-varying EMI that occurs during the actual acquisition.

[0066] Some dynamic interference estimation methods may utilize external sensors to detect EMI simultaneously during image acquisition, calculating linear transfer functions to predict and subtract EMI from the primary receive coils. However, these methods may be limited to k-spaceWSGR Docket No. 49880-719601implementations where multiple linear convolution models are built for consecutive phase-encode lines with correlated noise properties. Consequently, these methods may rely heavily on standard Cartesian k-space structures and Fourier sampling patterns. These methods can be generally incompatible with sparse sampling, undersampling techniques, unconventional nonCartesian imaging techniques, and systems utilizing heterogeneous magnetic fields or nonlinear encoding trajectories. Furthermore, these direct-fitting methods can often suffer from overfitting; because the auxiliary detectors capture random thermal noise in addition to deterministic EMI, directly fitting the full noise data inadvertently models and injects thermal noise into the primary MR signal, degrading the overall SNR.

[0067] Recent deep learning methods have been proposed to bypass linear transfer functions by learning complex, non-linear relationships between EMI sensing probes and primary receive coils. However, these models traditionally require massive training datasets, which can be exceptionally difficult to curate, particularly for unshielded environments. Utilizing simulated or high-field data to train these networks may result in severe domain shifts, and insufficiently representative training data can lead to image hallucinations when deployed clinically.

[0068] This problem can be particularly troublesome in low-field, single-sided, and portable MRI architectures that operate in unshielded environments and cannot rely on restrictive Faraday cages. Active EMI suppression methods can be inadequate for these systems because they either rely on rigid Cartesian k-space assumptions, require physically prohibitive sensor arrays, or depend on massive deep learning training datasets.

[0069] Therefore, there is a significant need for an improved EMI mitigation method that can function accurately across all MRI architectures, dynamically isolate true deterministic EMI from random detector noise without relying on rigid k-space trajectory assumptions and operate effectively without requiring massive training datasets or prohibitive sensor arrays.

[0070] To overcome these limitations, the disclosed method may apply dimensionality reduction to EMI data from auxiliary EMI sensors to extract a low rank noise subspace. This method may dynamically isolate EMI without relying on rigid k-space assumptions. By subtracting this isolated noise estimate, the system may effectively mitigate EMI and preserve the intrinsic signal -to-noise ratio across any spatial encoding trajectory or hardware configuration.

[0071] Provided herein are systems and methods for mitigating electromagnetic interference (EMI) in magnetic resonance (MR) imaging. In contrast to direct-fitting noise cancellation methods that often inadvertently inject thermal noise into the corrected signal, the disclosed techniques may utilize dimensionality reduction and subspace extraction. By isolating a low-rank approximation of the EMI, the present methods may effectively separate true deterministicWSGR Docket No. 49880-719601interference from random detector noise, thereby preserving the intrinsic Signal-to-Noise Ratio (SNR) of the MR data.

[0072] In some embodiments, the MR system acquires both primary MR signal data and auxiliary EMI data. In some embodiments, the primary MR signal data is acquired using one or more primary radiofrequency (RF) receiver coils. In some embodiments, simultaneously, or during designated noise-only acquisition windows within the pulse sequence, EMI data is obtained using at least one EMI detector. In some embodiments, the EMI detector may be positioned external to the MR imaging volume or integrated within the system. Examples of suitable EMI detectors may include, but are not limited to, E-field probes, auxiliary surface coils, conductive electrodes, and 50 Ohm terminated channels. In some embodiments, the primary radiofrequency (RF) receiver coils (Rx coils) themselves are utilized to collect EMI data during noise-only acquisition windows or known noise regions within the pulse sequence, either alone or in conjunction with auxiliary EMI detectors.

[0073] In some cases, the MR system utilizes between 1 and 32 EMI detectors to acquire auxiliary EMI data. In some cases, the number of EMI detectors is selected based on the dimensionality of the noise environment and the number of dominant EMI sources present at the installation site.

[0074] In some cases, the at least one EMI detector comprises one or more auxiliary surface coils constructed from electrically conductive materials including, but not limited to, copper, aluminum, silver, silver paste, litz wire, or any high electrical conducting material, including metal, alloys, or superconducting materials. In some cases, the auxiliary surface coils comprise loop coils, figure-8 coils, butterfly coils, or dipole antenna configurations. In some cases, each auxiliary surface coil has a diameter between about 10 mm and about 500 mm. In some cases, each auxiliary surface coil has a diameter between about 10 mm and about 50 mm, between about 30 mm and about 100 mm, between about 50 mm and about 200 mm, or between about 100 mm and about 500 mm, inclusive of any diameter therebetween.

[0075] In some cases, the at least one EMI detector comprises one or more auxiliary solenoids constructed from electrically conductive materials including, but not limited to, copper, aluminum, silver, silver paste, litz wire, or any high electrical conducting material, including metal, alloys, or superconducting materials. In some cases, each auxiliary solenoid has a diameter between about 1 mm and about 500 mm. In some cases, each auxiliary solenoid has a diameter between about 1 mm and about 50 mm, between about 30 mm and about 100 mm, between about 50 mm and about 200 mm, or between about 100 mm and about 500 mm, inclusive of any diameter therebetween.WSGR Docket No. 49880-719601

[0076] In some cases, the at least one EMI detector comprises a set of EMI sensor probes arranged in a mutually orthogonal configuration within an auxiliary enclosure. In some cases, the auxiliary enclosure houses three EMI sensor probes oriented along three orthogonal axes (X, Y, and Z) to capture the full vector components of the environmental EMI field. In some cases, the auxiliary enclosure is positioned external to the MR imaging volume and in proximity to the MR scanner housing. In some cases, the auxiliary enclosure comprises a non-conductive, nonmagnetic material such as a polymer, ceramic, or composite material. In some cases, the auxiliary enclosure has exterior dimensions between about 3 mm and about 300 mm along each axis, inclusive of any dimension therebetween.

[0077] In some cases, each EMI detector is tuned to a center frequency corresponding to the Larmor frequency of the MR system or to a frequency band associated with a known dominant EMI source. In some cases, each EMI detector is sensitive to frequencies between about 10 kHz and about 300 MHz. In some cases, each EMI detector is sensitive to frequencies between about 10 kHz and about 10 MHz, between about 100 kHz and about 5 MHz, between about 500 kHz and about 3 MHz, between about 1 MHz and about 10 MHz, or between about 10 MHz and about 300 MHz, inclusive of any frequencies therebetween. In some cases, the bandwidth of each EMI detector is between about 1 kHz and about 500 kHz, between about 5 kHz and about 200 kHz, or between about 10 kHz and about 100 kHz, inclusive of any bandwidth therebetween.

[0078] In some cases, each EMI detector has an input impedance of about 50 Ohms. In some cases, each EMI detector has an input impedance between about 10 Ohms and about 200 Ohms, between about 25 Ohms and about 100 Ohms, or between about 40 Ohms and about 75 Ohms, inclusive of any impedance therebetween. In some cases, each EMI detector is connected to a low-noise preamplifier and digitized by the MR spectrometer simultaneously with the primary MR receive channels. In some cases, the EMI detector signals are digitized at a sampling rate between about 100 kHz and about 100 MHz, between about 500 kHz and about 50 MHz, or between about 1 MHz and about 20 MHz, inclusive of any sampling rate therebetween.

[0079] In some cases, the at least one EMI detector comprises an E-field probe constructed from a short conductive element, such as a wire or rod antenna, with a length between about 5 mm and about 100 mm, enclosed in a non-conductive housing. In some cases, the at least one EMI detector comprises a conductive electrode configured to be attached to the patient’s skin using a biocompatible adhesive or conductive gel. In some cases, the at least one EMI detector comprises a 50 Ohm terminated channel on the MR spectrometer, wherein the terminated channel captures EMI coupled into the receive electronics without an external antenna element.WSGR Docket No. 49880-719601

[0080] In some embodiments, the EMI mitigation is performed as a pre-processing step prior to image reconstruction. In some embodiments, the system constructs a multi-dimensional noise data representation from the acquired EMI data to capture the temporal dynamics of the interference. In some cases, a sliding window of size k is applied across the zero-padded EMI data to generate a block Hankel convolution matrix, E.

[0081] In some cases, the sliding window size k is between about 2 and about 1024 samples. In some cases, the sliding window size k is between about 2 and about 16, between about 8 and about 64, between about 16 and about 128, between about 32 and about 256, between about 64 and about 512, or between about 128 and about 1024 samples, inclusive of any window size therebetween. In some cases, the sliding window size k is selected based on the temporal correlation length of the dominant EMI sources and the sampling rate of the EMI acquisition. In some embodiments, larger values of k capture longer-duration temporal correlations in the EMI signal at the cost of increased computational complexity.

[0082] In some cases, the block Hankel convolution matrix E has dimensions of M rows by N columns, where M corresponds to the number of time-shifted EMI observations and N corresponds to the product of the number of EMI detector channels and the sliding window size k. In some cases, the number of rows M is between about 100 and about 100,000, between about 500 and about 50,000, or between about 1,000 and about 10,000, inclusive of any value therebetween. In some cases, zero-padding is applied to the EMI data prior to constructing the block Hankel matrix to ensure consistent matrix dimensions across repetition time (TR) intervals of varying length.

[0083] In some embodiments, to prevent overfitting to the random thermal noise inherent in the EMI detectors, the system extracts a low-rank noise subspace from the convolution matrix. In some embodiments, the convolution matrix is subjected to a dimensionality reduction technique, such as singular value decomposition (SVD), represented as:Ecollapsed ~ U Vwhere L contains the singular values. In some embodiments, a truncated basis set, Vtrunc, is generated by retaining only the singular vectors corresponding to singular values exceeding a fractional threshold, e.g. st> T ■ smaxT i

[0084] In some cases, the predefined fractional threshold T is between about 0.001 and about 0.9. In some cases, the predefined fractional threshold T is between about 0.001 and about 0.01, between about 0.005 and about 0.05, between about 0.01 and about 0.1, between about 0.05 and about 0.2, or between about 0.1 and about 0.9, inclusive of any threshold value therebetween. In some cases, smaller values of T result in a larger truncated basis set that captures more noiseWSGR Docket No. 49880-719601components, while larger values of r produce a more compact basis set that retains only the most dominant EMI sources.

[0085] In some cases, the number of singular vectors retained in the truncated basis set is between about 1 and about 100. In some cases, the number of retained singular vectors is between about 1 and about 5, between about 2 and about 10, between about 5 and about 20, between about 10 and about 50, or between about 20 and about 100, inclusive of any number therebetween. In some cases, the effective rank of the noise subspace is determined automatically from the rate of decay of the singular values, without requiring manual specification by the operator. In some cases, the effective rank is between about 1 and about 50, between about 2 and about 20, or between about 3 and about 15 for an unshielded low-field MRI environment, inclusive of any rank therebetween.

[0086] In some embodiments, the multi-dimensional noise data representation is then projected onto this truncated basis set to form a reduced-dimensionality representation, E. In some embodiments, for each repetition time (TR) interval t and each primary MR signal channel p, a dynamic transformation is computed to map this reduced-dimensionality noise to the acquired MR signal Sp t. In some embodiments, by computing the transformation individually for each TR, the system accurately tracks dynamic, non-stationary environmental noise. In some embodiments, the transformation coefficients c may be calculated using a least-squares fit: c=

[0087] In some embodiments, a noise estimate Np t= Etc, which is subsequently subtracted from the original MR signal data to yield the corrected MR signal data, S'pt = Sp t— Np t. In some embodiments, the corrected MR signal data is then passed to a reconstruction pipeline to generate the final MR image.

[0088] In some cases, however, the measured MR receive signals contain both MRI signal and EMI noise. In some cases, estimating EMI statistics directly from signal-bearing data is therefore challenging. In some cases, the methods described herein address this challenge by exploiting structural properties of MRI signals across multiple receive coils.Signal and Noise Model

[0089] In some embodiments, the MRI signal received by each coil is related to the underlying image through the spatial sensitivity profile of that coil: Sp(r) = Cp(r) • x(r), where r denotes the spatial coordinate of a voxel or pixel within the reconstructed MRI image, Cprepresents the spatial sensitivity profile of coil p, and x represents the underlying MRI image.WSGR Docket No. 49880-719601

[0090] In some embodiments, because the coil sensitivity profiles determine the relative signal amplitude and phase received by each coil for a given spatial location, the ratio of MRI signals measured across different receive channels is determined by the corresponding coil sensitivity profiles. In some embodiments, any deviation from the expected signal ratios across coils therefore indicates the presence of noise, EMI contamination, or errors in the estimated coil sensitivity profiles.

[0091] In some embodiments, because the coil sensitivity profiles vary smoothly across space and are known or can be estimated, the MRI image can be reconstructed from the multi-coil data using standard reconstruction techniques. In some embodiments, once an MRI image estimate is obtained, it may be used together with the coil sensitivity profiles Cpto estimate the MRI signal contribution in each receive channel.

[0092] In some embodiments, an estimate of the EMI noise obtained by subtracting the estimated MRI signal S’pfrom the measured data, Np= Sp— Sp, may then be used to compute noise correlation matrices, convolution models, or other statistical quantities used in EMI cancellation algorithms.

[0093] In some embodiments, the EMI estimates obtained after signal subtraction may be used to construct or update the noise convolution matrix used in the SVD-based EMI cancellation method described. In some embodiments, this allows the convolution model to be estimated using data obtained from signal-bearing portions of the MRI acquisition rather than relying solely on noise-only calibration data.Iterative Joint Estimation of Image and Noise

[0094] In some embodiments, the noise estimation and subtraction steps are integrated directly into an iterative image reconstruction process. In some cases, this approach is advantageous because it prevents the algorithm from mathematically entangling true structural tissue signals (such as aliasing artifacts) with the EMI noise.

[0095] In some embodiments, the iterative reconstruction minimizes a joint objective function that simultaneously estimates the reconstructed MR image, x, and the noise estimate matrix, N . In some embodiments, the objective function may be expressed as:argmin- |y — (Ex + TV) I2 + AiR(x) + A2|7V |*X,N2where y represents the acquired MR signal data (k-space data), E is the spatial encoding operator, and R(x) is an image regularization term (e.g. Total Variation or wavelet sparsity) weighted by Xi.WSGR Docket No. 49880-719601

[0096] In some embodiments, the term || V ||* denotes the nuclear norm of the noise matrix, weighted by parameter A2. In some embodiments, the nuclear norm serves as a convex surrogate for the rank of the matrix, fundamentally enforcing the low-rank subspace constraint on the extracted EMI. In some embodiments, the acquired k-space data y may be either fully sampled or undersampled. In some embodiments, if fully sampled, the joint optimization cleanly separates the deterministic, low-rank EMI from the underlying tissue signal. In some embodiments, if undersampled, the optimization simultaneously performs parallel imaging / compressed sensing reconstruction utilizing ?(%) to mitigate aliasing, while the nuclear norm strictly isolates the interference.

[0097] In some embodiments, the system may execute an alternating minimization algorithm to solve this joint objective function, such as the Alternating Direction Method of Multipliers (ADMM) or a Primal-Dual algorithm. In some embodiments, the algorithm iteratively alternates between an image update step (updating x while holding N constant) and a noise update step (updating N while holding x constant). In some embodiments, during the noise update step, the system dynamically updates the truncated basis set by applying Singular Value Thresholding (SVT) to the data residual, thereby continuously refining the low-rank EMI approximation at each iteration until convergence is achieved.

[0098] In some cases, convergence of the alternating minimization algorithm is determined when a residual metric falls below a predefined convergence threshold. In some cases, the residual metric comprises a normalized difference between successive estimates of the reconstructed MR image, the noise estimate, or both. In some cases, the convergence threshold is between about 0.01% and about 10% of an initial residual value. In some cases, the convergence threshold is between about 0.01% and about 0.1%, between about 0.1% and about 1%, between about 0.5% and about 5%, or between about 1% and about 10%, inclusive of any threshold value therebetween.

[0099] In some cases, the alternating minimization algorithm is executed for a maximum number of iterations between about 1 and about 1000. In some cases, the maximum number of iterations is between about 1 and about 10, between about 5 and about 50, between about 10 and about 100, between about 50 and about 500, or between about 100 and about 1000, inclusive of any number of iterations therebetween. In some cases, the algorithm terminates when either the convergence threshold is reached or the maximum number of iterations is exhausted, whichever occurs first.WSGR Docket No. 49880-719601Iterative Joint Estimation of Signal and Noise

[0100] In some embodiments, the estimation of MRI signal and EMI noise may be performed iteratively, forming a joint estimation framework for separating signal and interference. In some embodiments, a procedure may include the following steps: (1) initial EMI suppression is performed using external EMI detectors or noise-only data; (2) the partially corrected multi-coil data are reconstructed into an MRI image; (3) the reconstructed image is used together with coil sensitivity profiles to estimate the MRI signal contribution for each receive channel; (4) the estimated signal contributions are subtracted from the measured receive channel signals; (5) the resulting residual signals provide improved estimates of EMI noise; and (6) the EMI cancellation model is updated using the improved noise estimates. In some embodiments, this process may be repeated until the signal and noise estimates converge.

[0101] In some cases, the iterative joint estimation procedure is repeated for between about 1 and about 50 iterations. In some cases, the number of iterations is between about 1 and about 5, between about 2 and about 10, between about 5 and about 20, or between about 10 and about 50, inclusive of any number therebetween. In some cases, convergence is determined when the change in the estimated noise correlation matrix between successive iterations falls below a predefined threshold, such as between about 0.1% and about 5% of the Frobenius norm of the noise correlation matrix.

[0102] In some embodiments, the MRI signal estimate uses information from all receive coils simultaneously, so it may have a significantly higher signal-to-noise ratio than the signal measured in any individual receive channel. In some cases, as a result, the subtraction of the estimated signal produces more accurate estimates of the EMI noise.Estimation of Coil Sensitivity Profiles

[0103] In some embodiments, the coil sensitivity profiles used in the signal estimation step may be obtained from calibration scans, electromagnetic simulations, or previously measured sensitivity maps. In some embodiments, the coil sensitivity profiles may be jointly estimated or refined together with the MRI signal and EMI noise.

[0104] In some embodiments, due to coil sensitivity profile variation smoothly across space, neighboring pixels or voxels may be used to improve the estimation of these profiles. In some embodiments, spatial smoothing, averaging, or local subspace estimation techniques may be used to improve sensitivity estimates. In some embodiments, joint estimation of MRI signal, coil sensitivity profiles, and EMI noise allows the algorithm to exploit additional structuralWSGR Docket No. 49880-719601information in the data and improves the robustness of EMI separation even when EMI signals are non-stationary.Method Overview

[0105] The methods described herein may suppress electromagnetic interference (EMI) in MRI receive signals by exploiting the structured relationship between MRI signals received by multiple coils and the statistical properties of EMI noise. In a multi-coil MRI system, signals measured by each receive coil may contain both MRI signal components and EMI noise components. In some embodiments, the MRI signal components across the receive channels are related through coil sensitivity profiles and therefore exhibit a predictable spatial relationship across the receive array. In some embodiments, this structural relationship allows an MRI image to be reconstructed from the multi-coil data and enables estimation of the MRI signal contribution in each individual receive channel.

[0106] In some embodiments, an MRI image is first reconstructed from the multi-coil data. In some embodiments, using coil sensitivity profiles, the reconstructed image is then used to estimate the MRI signal contribution in each receive channel. In some embodiments, these estimated signal contributions are subtracted from the measured data to obtain improved estimates of the EMI noise present in the receive channels. In some embodiments, the estimated EMI noise signals may then be used to compute noise correlation matrices, convolution models, or other statistical quantities used for EMI cancellation. In some cases, singular value decomposition (SVD) may be applied to a noise convolution matrix to identify dominant EMI components and generate a compressed basis for EMI estimation.

[0107] In some embodiments, the signal estimation, signal subtraction, and EMI cancellation steps are repeated iteratively. In some embodiments, each iteration improves the estimates of both the MRI signal and the EMI noise, allowing more accurate suppression of interference even when EMI signals are non-stationary or when noise-only calibration data is limited. In some embodiments, by jointly exploiting the spatial structure of MRI signals across receive coils and the statistical properties of EMI noise, the methods described herein enable improved separation of MRI signal and EMI interference in a wide range of MRI acquisition schemes.

[0108] In some embodiments, the system utilizes a historical noise subspace model to reduce computational latency and monitor environmental integrity. In some embodiments, the system aggregates EMI data over time to build a site-specific, longitudinal basis set, rather than constructing the full multi-dimensional noise data representation and computing the singular value decomposition (SVD) strictly during a live patient scan.WSGR Docket No. 49880-719601

[0109] In some embodiments, data for this historical model is systematically acquired during routine quality control (QC) protocols, which may be performed on a daily, weekly, or monthly basis. In some embodiments, phantom-based EMI measurements are recorded during these QC scans and appended to a historical database. In some embodiments, the system periodically updates the historical noise subspace model by applying the low-rank approximation (e.g., SVD and fractional thresholding) to the accumulated longitudinal data. In some embodiments, this generates a highly robust, pre-computed truncated basis set that characterizes the baseline electromagnetic environment of the specific installation site.

[0110] In some embodiments, during a subsequent patient scan, this pre-computed basis set can be utilized to bypass the computationally expensive steps of real-time matrix construction and SVD. In some embodiments, the system receives the live EMI data and projects it directly onto the historical basis set to estimate and subtract the baseline environmental interference. In some embodiments, a two-pass estimation is employed: the first pass utilizes the historical basis set to remove persistent baseline EMI, while a second, computationally lighter pass evaluates the residual noise to identify and remove transient interference specific to the current scan.[OHl] In some embodiments, the longitudinal noise data serves as an automated quality assurance (QA) diagnostic tool for the MR imaging environment. In some embodiments, the system executes a comparison between the noise subspace parameters of the current scan and the historical baseline model. In some cases, the system may calculate a deviation metric based on a shift in the singular value distribution or a sudden increase in the required rank to reach the fractional threshold. In some embodiments, if the deviation metric exceeds a predetermined threshold, it indicates the introduction of an anomalous broadband noise source or a degradation in the RF shielding of the MR scanner room. In some embodiments, the system triggers an automated action in response, such as generating a user alert, automatically increasing the number of signal averages (NEX) to compensate for the elevated noise floor, or switching to an alternative noise cancellation protocol.

[0112] In some embodiments, the dimensionality reduction, subspace extraction, and iterative reconstruction steps are performed using trained artificial neural networks. In some embodiments, these deep learning implementations serve as functional equivalents to the linear algebraic operations (e.g., SVD and nuclear norm regularization) described in prior embodiments, offering potential advantages in inference speed and adaptive feature extraction.

[0113] In some embodiments, the extraction of the noise subspace is performed using an undercomplete autoencoder. In some embodiments, the multi-dimensional noise data representation (such as the block Hankel matrix) is provided as input to the autoencoder. In someWSGR Docket No. 49880-719601embodiments, the autoencoder comprises an encoder network that compresses the input data into a lower-dimensional latent space at a bottleneck layer, and a decoder network that reconstructs the data from the latent space. In some embodiments, by constraining the dimensionality of the bottleneck layer, the network is forced to learn a low-rank approximation of the input data. In some embodiments, the latent representation output by the bottleneck layer thereby serves as the truncated basis set, isolating the deterministic EMI components while filtering out random thermal noise. In some embodiments, the autoencoder is trained using a loss function (e.g., mean squared error) that minimizes the reconstruction error over a dataset of historical, site-specific EMI data.

[0114] In some embodiments, the alternating minimization algorithm (such as ADMM or Primal-Dual) is mapped to a deep learning architecture using algorithm unrolling. In some embodiments, in an unrolled neural network, each iteration of the optimization algorithm is represented as a distinct layer or block within the network. In some embodiments, this allows the regularization parameters (2) and the thresholding operators to be learned directly from data rather than being manually tuned.

[0115] In some embodiments, specifically, the noise update step involving Singular Value Thresholding (SVT) is parameterized as a non-linear activation function within the network layers, utilizing learnable threshold values. In some embodiments, the unrolled network takes the EMI-corrupted k-space data and the auxiliary EMI data as inputs and outputs the corrected, unaliased MR image. In some embodiments, the network is trained via supervised learning using a dataset containing pairs of retrospectively or prospectively acquired EMI-corrupted MR k-space data and clean, ground-truth unaliased MR images. In some embodiments, the training process updates the network weights by minimizing a loss function, such as an fxor f2norm, calculating the difference between the network output and the ground-truth images.

[0116] While the methods described herein can be fully applicable to fully shielded, high-field MRI systems, they may provide distinct advantages for specialized hardware. In some embodiments, the primary MR signal data is acquired using a low-field or ultra-low-field MR scanner, a single-sided MR scanner, or a portable MR scanner configured for point-of-care operation. In some embodiments, because the present subspace method efficiently isolates dominant EMI components without requiring rigid Cartesian sampling, it enables high-fidelity imaging in unshielded environments lacking stationary Faraday cages and functions robustly despite the restricted sensor placement geometries inherent to open or single-sided MRI systems.

[0117] In some embodiments, the MRI system contains a BO field that is non-uniform along the z-direction (away from the system) and therefore operates over a much larger relative bandwidthWSGR Docket No. 49880-719601than traditional MRI systems that operate in a uniform BO magnetic field with a predominantly single operating frequency. In some embodiments, to cover this entire bandwidth, the data acquisition method employs multiple excited bands, each with a center frequency and bandwidth so as to cover the entire system bandwidth associated with the desired z-FOV.

[0118] In some embodiments, EMI noise sources tend to have bandwidths that are significantly smaller than the bandwidth of an excited band. In some embodiments, some noise coils may be tuned to a few strong EMI sources while other noise coils are tuned to cover the entire bandwidth of a slab. In some embodiments, this allows the noise cancellation method to efficiently target both narrowband and broadband interference components within the system’s operating bandwidth.

[0119] In some cases, the EMI mitigation methods described herein are implemented by the processing module and spectrometer of the MR imaging system. In some embodiments, the spectrometer simultaneously digitizes both the primary MR receive channel signals and the auxiliary EMI detector signals. In some embodiments, the processing module receives the digitized data and executes the noise convolution matrix construction, dimensionality reduction, dynamic transformation, and noise subtraction operations described herein. In some cases, the programmable logic controller (PLC) of the MR system manages the timing of EMI data acquisition windows within the pulse sequence, including enabling and disabling EMI detector channels during designated noise-only periods and during signal acquisition periods.

[0120] In some cases, the auxiliary EMI detectors are physically mounted to or proximate to the MR scanner housing, external to the imaging volume and the gradient coil set. In some cases, the EMI detector signals are routed through dedicated receive channels on the spectrometer, separate from the primary MR receive coil channels. In some cases, the spectrometer digitizes the EMI detector channels using the same clock source and timing reference as the primary MR receive channels to ensure phase coherency between EMI and MR signal measurements.Permanent Magnet

[0121] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can include a permanent magnet.

[0122] FIGS. 8A-8B is a schematic illustration of a magnetic resonance imaging system 800, In some cases. The system 800 includes a housing 820. As shown in FIGS. 8A-8B, the housing 820 includes a permanent magnet 830, a radio frequency transmit coil 840, a gradient coil set 850, an optional electromagnet 860, a radio frequency receive coil 870, and a power source 880.In some cases, the system 800 can include various electronic components, such as for example,WSGR Docket No. 49880-719601but not limited to a varactor, a PIN diode, a capacitor, or a switch, including a micro-electro-mechanical system (MEMS) switch, a solid-state relay, or a mechanical relay. In some cases, the various electronic components listed above can be configured with the radio frequency transmit coil 840.

[0123] FIG. 8A is a schematic illustration of a magnetic resonance imaging system 800, In some cases. FIG. 8B illustrates an exploded view of the magnetic resonance imaging system 800. FIG. 8C is a schematic front view of the magnetic resonance imaging system 800, In some cases. FIG. 8D is a schematic side view of the magnetic resonance imaging system 800, In some cases. As shown in FIG. 8A and FIG. 8B, the magnetic resonance imaging system 800 includes a housing 820. The housing 820 includes a front surface 825. In some cases, the front surface 825 can be a concave front surface. In some cases, the front surface 825 can be a recessed front surface.

[0124] In some cases, the permanent magnet 830 provides a static magnetic field in a region of interest 890 (also referred to herein as "given field of view"). In some cases, the permanent magnet 830 can include a plurality of cylindrical permanent magnets in parallel configuration as shown in FIG. 8C and FIG. 8D. In some cases, the permanent magnet 830 can include any suitable magnetic materials, including but not limited, to rare-earth based magnetic materials, such as for example, Nd-based magnetic materials, and the like. As shown in FIG. 8A, the main permanent magnet might include an access aperture 835 for accessing the patient from multiple sides of the system.

[0125] In some cases, the static magnetic field of the permanent magnet 230 may vary from about 50 mT to about 60 mT, about 45 mT to about 65 mT, about 40 mT to about 70 mT, about 35 mT to about 75 mT, about 30 mT to about 80 mT, about 25 mT to about 85 mT, about 20 mT to about 90 mT, about 15 mT to about 95 mT and about 10 mT to about 100 mT to a given field of view. The magnetic field may also vary from about 10 mT to about 15 mT, about 15 mT to about 20 mT, about 20 mT to about 25 mT, about 25 mT to about 30 mT, about 30 mT to about 35 mT, about 35 mT to about 40 mT, about 40 mT to about 45 mT, about 45 mT to about 50 mT, about 50 mT to about 55 mT, about 55 mT to about 60 mT, about 60 mT to about 65 mT, about 65 mT to about 70 mT, about 70 mT to about 75 mT, about 75 mT to about 80 mT, about 80 mT to about 85 mT, about 85 mT to about 90 mT, about 90 mT to about 95 mT, and about 95 mT to about 100 mT. In some cases, the static magnetic field of the permanent magnet 230 may also vary from about 1 mT to about 1 T, about 10 mT to about 195 mT, about 15 mT to about 900 mT, about 20 mT to about 800 mT, about 25 mT to about 700 mT, about 30 mT to about 600 mT, about 35 mT to about 500 mT, about 40 mT to about 400 mT, about 45 mT to about 300WSGR Docket No. 49880-719601mT, about 50 mT to about 200 mT, about 50 mT to about 100 mT, about 45 mT to about 100 mT, about 40 mT to about 100 mT, about 35 mT to about 100 mT, about 30 mT to about 100 mT, about 25 mT to about 100 mT, about 20 mT to about 100 mT, and about 15 mT to about 100 mT.

[0126] In some cases, the permanent magnet 830 can include a bore 835 in its center. In some cases, the permanent magnet 830 may not include a bore. In some cases, the bore 835 can have a diameter between 1 inch and 20 inches. In some cases, the bore 835 can have a diameter between 1 inch and 4 inches, between 4 inches and 8 inches, and between 10 inches and 20 inches. In some cases, the given field of view can be a spherical or cylindrical field of view, as shown in FIG. 8A and FIG. 8B. In some cases, the spherical field of view can be between 2 inches and 20 inches in diameter. In some cases, the spherical field of view can have a diameter between 1 inch and 4 inches, between 4 inches and 8 inches, and between 10 inches and 20 inches. In some cases, the cylindrical field of view is approximately between 2 inches and 20 inches in length. In some cases, the cylindrical field of view can have a length between 1 inch and 4 inches, between 4 inches and 8 inches, and between 10 inches and 20 inches.

[0127] In some cases, the permanent magnet system can be rotated 90 degrees such that the system has an imaging field of view on top of the system. This field of view can allow a patient to sit down on top of the system so that the biological material can be imaged with gravity affecting the tissues and structures.Radio Frequency Transmit Coil

[0128] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a radio frequency transmit coil.

[0129] FIG. 9 is a schematic view of an implementation of a magnetic imaging apparatus 900, according to various embodiments. As shown in FIG. 9, the apparatus 900 includes a radio frequency transmit coil 920 that projects the RF power outwards away from the coil 920. The coil 920 has two rings 922 and 924 that are connected by one or more rungs 926. As shown in FIG. 9, the coil 920 is also connected to a power source 950a and / or a power source 950b (collectively referred to herein as "power source 950"). In some cases, power sources 950a and 950b can be configured for power input and / or signal input and can be referred to as coil input. In some cases, the power source 950a and / or 950b are configured to provide contact via electrical contacts 952a and / or 952b (collectively referred to herein as "electrical contact 952"), and electrical contacts 954a and / or 954b (collectively referred to herein as "electrical contactWSGR Docket No. 49880-719601954") by attaching the electrical contacts 952 and 954 to one or more rungs 926. The coil 920 is configured to project a uniform RF field within a field of view 940. In some cases, the field of view 940 is a region of interest for magnetic resonance imaging (i.e., imaging region) where a patient resides. Since the patient resides in the field of view 940 away from the coil 920, the apparatus 900 is suitable for use in a single-sided magnetic resonance imaging system. In some cases, the coil 920 can be powered by two signals that are 90 degrees out of phase from each other, for example, via quadrature excitation.

[0130] In some cases, the coils can be comprised of two or more orthogonal figure-8 shapes arranged on the surface of the magnetic surface. These coils can then be tuned to the same or to different RF resonant frequencies. These coils are designed to generate a uniform or varying magnetic RF field within the region of interest that is off-of the face of the magnet.

[0131] In some cases, the coil 920 includes the ring 922 and the ring 924 that are positioned coaxially along the same axis but at a distance away from each other, as shown in FIG. 9. In some cases, the ring 922 and the ring 924 are separated by a distance ranging from about 0.1 m to about 10 m. In some cases, the ring 922 and the ring 924 are separated by a distance ranging from about 0.2 m to about 5 m, about 0.3 m to about 2 m, about 0.2 m to about 1 m, about 0.1 m to about 0.8 m, or about 0.1 m to about 1 m, inclusive of any separation distance therebetween. In some cases, the coil 920 includes the ring 922 and the ring 924 that are positioned non-co-axially but along the same direction and separated at a distance ranging from about 0.2 m to about Sm. In some cases, the ring 922 and the ring 924 can also be tilted with respect to each other. In some cases, the tilt angle can be from 1 degree to 90 degrees, from 1 degree to 5 degrees, from 5 degrees to 10 degrees, from 10 degrees to 25 degrees, from 25 degrees to 45 degrees, and from 45 degrees to 90 degrees.

[0132] In some cases, the ring 922 and the ring 924 have the same diameter. In some cases, the ring 922 and the ring 924 have different diameters and the ring 922 has a larger diameter than the ring 924, as shown in FIG. 9. In some cases, the ring 922 and the ring 924 have different diameters and the ring 922 has a smaller diameter than the ring 924. In some cases, the ring 922 and the ring 924 of the coil 320 are configured to create the imaging region in the field of view 940 containing a uniform RF power profile within the field of view 940, a field of view that is not centered within the RF-TX coil and is instead projected outwards in space from the coil itself.

[0133] In some cases, the ring 922 has a diameter between about 10 pm and about 10 m. In some cases, the ring 922 has a diameter between about 0.001 m and about 9 m, between about 0.01 m and about 8 m, between about 0.03 m and about 6 m, between about 0.05 m and about 5 m,WSGR Docket No. 49880-719601between about 0.1 m and about 3 m, between about 0.2 m and about 2 m, between about 0.3 m and about 1.5 m, between about 0.5 m and about 1 m, or between about 0.01 m and about 3 m, inclusive of any diameter therebetween.

[0134] In some cases, the ring 924 has a diameter between about 10 gm and about 10 m. In some cases, the ring 924 has a diameter between about 0.001 m and about 9 m, between about 0.01 m and about 8 m, between about 0.03 m and about 6 m, between about 0.05 m and about 5 m, between about 0.1 m and about 3 m, between about 0.2 m and about 2 m, between about 0.3 m and about 1.5 m, between about 0.5 m and about 1 m, or between about 0.01 m and about 3 m, inclusive of any diameter therebetween.

[0135] In some cases, the ring 922 and the ring 924 are connected by one or more rungs 926, as shown in FIG. 9. In some cases, the one or more rungs 926 are connected to the ring 922 and 924 so as to form a single electrical circuit loop (or single current loop). As shown in FIG. 9, for example, one end of the one or more rungs 926 is connected to the electrical contact 952 of the power source 950 and another end of the one or more rungs 926 be connected to the electrical contact 954 so that the coil 920 completes an electrical circuit.

[0136] In some cases, the ring 922 is a discontinuous ring and the electrical contact 952 and the electrical contact 954 can be electrically connected to two opposite ends of the ring 922 to form an electrical circuit powered by the power source 950. Similarly, in some cases, the ring 924 is a discontinuous ring and the electrical contact 952 and the electrical contact 954 can be electrically connected to two opposite ends of the ring 924 to form an electrical circuit powered by the power source 950.

[0137] In some cases, the rings 922 and 924 are not circular and can instead have a cross section that is elliptical, square, rectangular, or trapezoidal, or any shape or form having a closed loop. In some cases, the rings 922 and 924 may have cross sections that vary in two different axial planes with the primary axis being a circle and the secondary axis having a sinusoidal shape or some other geometric shape. In some cases, the coil 920 may include more than two rings 922 and 924, each connected by rungs that span and connect all the rings. In some cases, the coil 920 may include more than two rings 922 and 924, each connected by rungs that alternate connection points between rings. In some cases, the ring 922 may contain a physical aperture for access. In some cases, the ring 922 may be a solid sheet without a physical aperture.

[0138] In some cases, the coil 920 generates an electromagnetic field (also referred to herein as "magnetic field") strength between about 1 pT and about 10 mT. In some cases, the coil 920 can generate a magnetic field strength between about 10 pT and about 5 mT, about 50 pT and about 1 mT, or about 100 pT and about 1 mT, inclusive of any magnetic field strength therebetween.WSGR Docket No. 49880-719601

[0139] In some cases, the coil 920 generates an electromagnetic field that is pulsed at a radio frequency between about 1 kHz and about 2 GHz. In some cases, the coil 920 generates a magnetic field that is pulsed at a radio frequency between about 1 kHz and about 1 GHz, about 10 kHz and about 800 MHz, about 50 kHz and about 300 MHz, about 100 kHz and about 100 MHz, about 10 kHz and about 10 MHz, about 10 kHz and about 5 MHz, about 1 kHz and about 2 MHz, about 50 kHz and about 150 kHz, about 80 kHz and about 120 kHz, about 800 kHz and about 1.2 MHz, about 100 kHz and about 10 MHz, or about 1 MHz and about 5 MHz, inclusive of any frequencies therebetween.

[0140] In some cases, the coil 920 is oriented to partially surround the region of interest. In some cases, the ring 922, the ring 924, and the one or more rungs 926 are non-planar to each other. Said another way, the ring 922, the ring 924, and the one or more rungs 926 form a three-dimensional structure that surrounds the region of interest where a patient resides. In some cases, the ring 922 is closer to the region of interest than the ring 924, as shown in FIG. 9. In some cases, the region of interest has a size of about 0.1 m to about 1 m. In some cases, the region of interest is smaller than the diameter of the ring 922. In some cases, the region of interest is smaller than both the diameter of the ring 924 and the diameter of the ring 922, as shown in FIG.9. In some cases, the region of interest has a size that is smaller than the diameter of the ring 922 and larger than the diameter of the ring 924.

[0141] In some cases, the ring 922, the ring 924, or the rungs 926 include the same material. In some cases, the ring 922, the ring 924, or the rungs 926 include different materials. In some cases, the ring 922, the ring 924, or the rungs 926 include hollow tubes or solid tubes. In some cases, the hollow tubes or solid tubes can be configured for air or fluid cooling. In some cases, each of the ring 922 or the ring 924 or the rungs 926 includes one or more electrically conductive windings. In some cases, the windings include litz wires or any electrical conducting wires. These additional windings can be used to improve performance by lowering the resistance of the windings at the desired frequency. In some cases, the ring 922, the ring 924, or the rungs 926 include copper, aluminum, silver, silver paste, or any high electrical conducting material, including metal, alloys or superconducting metal, alloys or non-metal. In some cases, the ring 922, the ring 924, or the rungs 926 may include metamaterials.

[0142] In some cases, the ring 922, the ring 924, or the rungs 926 may contain separate electrically non-conductive thermal control channels designed to maintain the temperature of the structure to a specified setting. In some cases, the thermal control channels can be made from electrically conductive materials and integrated as to carry the electrical current.WSGR Docket No. 49880-719601

[0143] In some cases, the coil 920 includes one or more electronic components for tuning the magnetic field. The one or more electronic components can include a varactor, a PIN diode, a capacitor, or a switch, including a micro-electro-mechanical system (MEMS) switch, a solid-state relay, or a mechanical relay. In some cases, the coil can be configured to include any of the one or more electronic components along the electrical circuit. In some cases, the one or more components can include mu metals, dielectrics, magnetic, or metallic components not actively conducting electricity and can tune the coil. In some cases, the one or more electronic components used for tuning includes at least one of dielectrics, conductive metals, metamaterials, or magnetic metals. In some cases, tuning the electromagnetic field includes changing the current or by changing physical locations of the one or more electronic components. In some cases, the coil is cryogenically cooled to reduce resistance and improve efficiency. In some cases, the first ring and the second ring comprise a plurality of windings or litz wires.

[0144] In some cases, the coil 920 is configured for a magnetic resonance imaging system that has a magnetic field gradient across the field of view. The field gradient allows for imaging slices of the field of view without using an additional electromagnetic gradient. As disclosed herein, the coil can be configured to generate a large bandwidth by combining multiple center frequencies, each with their own bandwidth. By superimposing these multiple center frequencies with their respective bandwidths, the coil 920 can effectively generate a large bandwidth over a desired frequency range between about 1 kHz and about 2 GHz. In some cases, the coil 920 generates a magnetic field that is pulsed at a radio frequency between about 10 kHz and about 800 MHz, about 50 kHz and about 300 MHz, about 100 kHz and about 100 MHz, about 10 kHz and about 10 MHz, about 10 kHz and about 5 MHz, about 1 kHz and about 2 MHz, about 50 kHz and about 150 kHz, about 80 kHz and about 120 kHz, about 800 kHz and about 1.2 MHz, about 100 kHz and about 10 MHz, or about 1 MHz and about 5 MHz, inclusive of any frequencies therebetween.Gradient Coil Set

[0145] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a gradient coil set.

[0146] FIG. 10 is a schematic view of an implementation of a magnetic imaging apparatus 1000, according to various embodiments. As shown in FIG. 10, the apparatus 1000 includes a gradient coil set 1020 (also referred to herein as single-sided gradient coil set 1020) that is configured to project a gradient magnetic field outwards away from the coil set 1020 and within a field of viewWSGR Docket No. 49880-7196011030. In some cases, the field of view 1030 is a region of interest for magnetic resonance imaging (i.e., imaging region) where a patient resides. Since the patient resides in the field of view 1030 away from the coil set 1020, the apparatus 1000 is suitable for use in a single-sided MRI system.

[0147] As shown in the figure, the coil set 1020 includes variously sized spiral coils in various sets of spiral coils 1040a, 1040b, 1040c, and 1040d (collectively referred to as "spiral coils 1040"). Each set of the spiral coils 1040 include at least one spiral coil and FIG. 10 is shown to include 3 spiral coils. In some cases, each spiral coil in the spiral coils 1040 has an electrical contact at its center and an electrical contact output on the outer edge of the spiral coil so as to form a single running loop of electrically conducting material spiraling out from the center to the outer edge, or vice versa. In some cases, each spiral coil in the spiral coils 1040 has a first electrical contact at a first position of the spiral coil and a second electrical contact at a second position the spiral coil so as to form a single running loop of electrically conducting material from the first position to the second position, or vice versa.

[0148] As shown in FIG. 10, the coil set 1020 also includes an aperture 1025 at its center where the spiral coils 1040 are disposed around the aperture 1025. The aperture 1025 itself does not contain any coil material within it for generating magnetic material. The coil set 1020 also includes an opening 1027 on the outer edge of the coil set 1020 to which the spiral coils 1040 can be disposed. Said another way, the aperture 1025 and the opening 1027 define the boundaries of the coil set 1020 within which the spiral coils 1040 can be disposed. In some cases, the coil set 1020 forms a bowl shape with a hole in the center.

[0149] In some cases, the spiral coils 1040 form across the aperture 1025. For example, the spiral coils 1040a are disposed across from the spiral coils 1040c with respect to the aperture 1025. Similarly, the spiral coils 1040b are disposed across from the spiral coils 1040d with respect to the aperture 1025. In some cases, the spiral coils 1040 in the coil set 1020 shown in FIG. 10 are configured to create spatial encoding in the magnetic gradient field within the field of view 1030.

[0150] As shown in FIG. 10, the coil set 1020 is also connected to a power source 450 via electrical contacts 1052 and 1054 by attaching the electrical contacts 1052 and 1054 to one or more of the spiral coils 1040. In some cases, the electrical contact 1052 is connected to one of the spiral coils 1040, which is then connected to other spiral coils 1040 in series and / or in parallel, and one other spiral coil 1040 is then connected to the electrical contact 1054 so as to form an electrical current loop. In some cases, the spiral coils 1040 are all electrically connected in series. In some cases, the spiral coils 1040 are all electrically connected in parallel. In someWSGR Docket No. 49880-719601cases, some of the spiral coils 1040 are electrically connected in series while other spiral coils 1040 are electrically connected in parallel. In some cases, the spiral coils 1040a are electrically connected in series while the spiral coils 1040b are electrically connected in parallel. In some cases, the spiral coils 1040c are electrically connected in series while the spiral coils 1040d are electrically connected in parallel. The electrical connections between each spiral coil in the spiral coils 1040 or each set of spiral coils 1040 can be configured as needed to generate the magnetic field in the field of view 1030.

[0151] In some cases, the coil set 1020 includes the spiral coils 1040 spread out as shown in FIG. 10. In some cases, each of the sets of spiral coils 1040a, 1040b, 1040c, and 1040d are configured in a line from the aperture 1025 to the opening 1027 so that each set of spiral coils is set apart from another by an angle of 90°. In some cases, 1040a and 1040b are set at 45° from one another, and 1040c and 1040d are set at 45° from one another, while 1040c is set 135° on the other side of 1040b and 1040d is set 135° on the other side of 1040a. In essence, any of the sets of spiral coils 1040 can be configured in any arrangement for any number "n" of sets of spiral coils 1040.

[0152] In some cases, the spiral coils 1040 have the same diameter. In some cases, each of the sets of spiral coils 1040a, 1040b, 1040c, and 1040d have the same diameter. In some cases, the spiral coils 1040 have different diameters. In some cases, each of the sets of spiral coils 1040a, 1040b, 1040c, and 1040d have different diameters. In some cases, the spiral coils in each of the sets of spiral coils 1040a, 1040b, 1040c, and 1040d have different diameters. In some cases, 1040a and 1040b have the same first diameter and 1040c and 1040d have the same second diameter, but the first diameter and the second diameter are not the same.

[0153] In some cases, each spiral coil in the spiral coils 1040 has a diameter between about 10 pm and about 10 m. In some cases, each spiral coil in the spiral coils 1040 has a diameter between about 0.001 m and about 9 m, between about 0.005 m and about 8 m, between about 0.01 m and about 6 m, between about 0.05 m and about 5 m, between about 0.1 m and about 3 m, between about 0.2 m and about 2 m, between about 0.3 m and about 1.5 m, between about 0.5 m and about 1 m, or between about 0.01 m and about 3 m, inclusive of any diameter therebetween.

[0154] In some cases, the spiral coils 1040 are connected to form a single electrical circuit loop (or single current loop). As shown in FIG. 10, for example, one spiral coil in the spiral coils 1040 is connected to the electrical contact 1052 of the power source 450 and another spiral coil is connected to the electrical contact 1054 so that the spiral coils 1040 completes an electrical circuit.WSGR Docket No. 49880-719601

[0155] In some cases, the coil set 1020 generates an electromagnetic field strength (also referred to herein as "electromagnetic field gradient" or "gradient magnetic field") between about 1 pT and about 10 T. In some cases, the coil set 1020 can generate an electromagnetic field strength between about 100 gT and about 1 T, about 1 mT and about 500 mT, or about 10 mT and about 100 mT, inclusive of any magnetic field strength therebetween. In some cases, the coil set 1020 can generate an electromagnetic field strength greater than about 1 pT, about 10 gT, about 100 pT, about 1 mT, about 5 mT, about 10 mT, about 20 mT, about 50 mT, about 100 mT, or about 500 mT.

[0156] In some cases, the coil set 1020 generates an electromagnetic field that is pulsed at a rate with a rise-time less than about 100 ps. In some cases, the coil set 1020 generates an electromagnetic field that is pulsed at a rate with a rise-time less than about 1 ps, about 5 ps, about 10 ps, about 20 ps, about 30 ps, about 40 ps, about 50 ps, about 100 ps, about 200 ps, about 500 ps, about 1 ms, about 2 ms, about 5 ms, or about 10 ms.

[0157] In some cases, the coil set 1020 is oriented to partially surround the region of interest in the field of view 1030. In some cases, the spiral coils 1040 are non-planar to each other. In some cases, the sets of spiral coils 1040a, 1040b, 1040c, and 1040d are non-planar to each other. Said another way, the spiral coils 1040 and each of the sets of spiral coils 1040a, 1040b, 1040c, and 1040d form a three-dimensional structure that surrounds the region of interest in the field of view 1030 where a patient resides.

[0158] In some cases, the spiral coils 1040 include the same material. In some cases, the spiral coils 1040 include different materials. In some cases, the spiral coils in set 1040a include the same first material, the spiral coils in set 1040b include the same second material, the spiral coils in set 1040c include the same third material, the spiral coils in set 1040d include the same fourth material, but the first, second, third and fourth materials are different materials. In some cases, the first and second materials are the same material, but that same material is different from the third and fourth materials, which are the same. In essence, any of the spiral coils 1040 can be of the same material or different materials depending on the configuration of the coil set 1020.

[0159] In some cases, the spiral coils 1040 include hollow tubes or solid tubes. In some cases, the spiral coils 1040 include one or more windings. In some cases, the windings include litz wires or any electrical conducting wires. In some cases, the spiral coils 1040 include copper, aluminum, silver, silver paste, or any high electrical conducting material, including metal, alloys or superconducting metal, alloys or non-metal. In some cases, the spiral coils 1040 include metamaterials.WSGR Docket No. 49880-719601

[0160] In some cases, the coil set 1020 includes one or more electronic components for tuning the magnetic field. The one or more electronic components can include a PIN diode, a mechanical relay, a solid-state relay, or a switch, including a micro electro-mechanical system (MEMS) switch. In some cases, the coil can be configured to include any of the one or more electronic components along the electrical circuit. In some cases, the one or more components can include mu metals, dielectrics, magnetic, or metallic components not actively conducting electricity and can tune the coil. In some cases, the one or more electronic components used for tuning includes at least one of conductive metals, metamaterials, or magnetic metals. In some cases, tuning the electromagnetic field includes changing the current or by changing physical locations of the one or more electronic components. In some implementations, the coil is cryogenically cooled to reduce resistance and improve efficiency.Electromagnet

[0161] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include an electromagnet.

[0162] FIG. 11 is a schematic front view of a magnetic resonance imaging system 1100, according to various embodiments. In some cases, the system 1100 can be any magnetic resonance imaging system, including for example, a single-sided magnetic resonance imaging system that comprises a magnetic resonance imaging scanner or a magnetic resonance imaging spectrometer, as disclosed herein.

[0163] As shown in FIG. 11, the system 1100 includes a housing 1120 that can house various components, including, for example but not limited to, magnets, electromagnets, coils for producing radio frequency fields, various electronic components, for example but not limited to, for controlling, powering, and / or monitoring of the system 1100. In some cases, the housing 1120 can house, for example, the permanent magnet 930, the radio frequency transmit coil 940, and / or the gradient coil set 950 within the housing 1120. In some cases, the system 1100 also includes a bore 1135 in its center. As shown in FIG. 11, the housing 1120 also includes a front surface 1125 of the system 1100. In some cases, the front surface 1125 can be curved, flat, concave, convex, or otherwise have a straight or curvilinear surface. In some cases, the magnetic resonance imaging system 1100 can be configured to provide a region of interest in field of view 1130

[0164] As shown in FIG. 11, the system 1100 includes an electromagnet 1160 disposed proximate to the front surface 1125 of the system 1100. In some cases, the electromagnet 1160 is disposed proximate to the center of the front surface 1125 on the front side of the system 1100.WSGR Docket No. 49880-719601In some cases, the electromagnet 1160 can be a solenoid coil configured to create a field that either adds or subtracts from the magnetic field, for example, of the permanent magnet 930. In some cases, this field can create a prepolarizing field for enhancing the signal or contrast from the nuclear magnetic resonance.

[0165] As shown in FIG. 11, the given field of view 1130 resides at the center of the front surface 1125 of the system 1100. In some cases, the electromagnet 1160 is disposed within the given field of view 1130. In some cases, the electromagnet 1160 is disposed concentrically with the given field of view 1130. In some cases, the electromagnet 1160 can be inserted in the bore 1135. In some cases, the electromagnet 1160 can be placed proximate to the bore 1135. For example, the electromagnet 1160 can be placed in front, back or middle of the bore 1135. In some cases, the electromagnet 1160 can be placed proximate to, or at the entrance of the bore 1135Radio Frequency Receive Coil

[0166] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a radio frequency receive coil.

[0167] Some MR systems can create a uniform field within the imaging region. This uniform field can then generate a narrow band of magnetic resonance frequencies that can then be captured by a receive coil, amplified, and digitized by a spectrometer. Since frequencies are within a narrow well-defined bandwidth, hardware architecture can be focused on creating a statically tuned RF-RX coil with an optimal coil quality factor. Many variations in coil architectures have been created that explore large single volume coils, coil arrays, parallelized coil arrays, or body specific coil arrays. However, these structures may be limited to imaging a specific frequency close to the region of interest at high field strengths and with a limited sized region of interest within a magnetic bore.

[0168] In some cases, an MRI system is provided that can include an imaging region that can be offset from a face of a magnet. The MRI system can provide relatively unobstructed imaging. The MRI system can have a built-in magnetic field gradient that creates a range of field values over the region of interest. The MRI system can operate at a lower magnetic field strength as compared to other MRI systems allowing for a relaxation on the RX coil design constraints and allowing for additional mechanisms like robotics to be used with the MRI.

[0169] The architecture of the main magnetic field of the MRI system, in some cases, can create a different set of optimization constraints. Because the imaging volume can extend over aWSGR Docket No. 49880-719601broader range of magnetic resonance frequencies, the hardware can be configured to be sensitive to and capture the specific frequencies that are generated across the field of view. This frequency spread can be much larger than a single receive coil tuned to a single frequency and can provide increased sensitivity. In addition, because the field strength can be much lower than other MRI systems, and because signal intensity can be proportional to the field strength, it can allow for a maximization of a signal to noise ratio of the receive coil network. Methods are therefore provided, in some cases, to acquire the full range of frequencies that are generated within the field of view without loss of sensitivity.

[0170] In some cases, several methods are provided that can allow for imaging within the MRI system. These methods can include combining: (i) a variable tuned RF-RX coil; (ii) a RF-RX coil array with elements tuned to frequencies that are dependent upon the spatial inhomogeneity of the magnetic field; (iii) an ultralow-noise pre-amplifier design; and (iv) an RF-RX array with multiple receive coils designed to optimize the signal from a defined and limited field of view for a specific body part. These methods can be combined in any combination as needed.

[0171] In some cases, a variable tuned RF-RX coil can comprise one or more electronic components for tuning the electromagnetic receive field. In some cases, the one or more electronic components can include at least one of a varactor, a PIN diode, a capacitor, an inductor, a MEMS switch, a solid-state relay, or a mechanical relay. In some cases, the one or more electronic components used for tuning can include at least one of dielectrics, capacitors, inductors, conductive metals, metamaterials, or magnetic metals. In some cases, tuning the electromagnetic receive field includes changing the current or by changing physical locations of the one or more electronic components. In some cases, the coil is cryogenically cooled to reduce resistance and improve efficiency.

[0172] In some cases, the RF-RX array can be comprised of individual coil elements that are each tuned to a variety of frequencies. The appropriate frequency can be chosen, for example, to match the frequency of the magnetic field located at the specific spatial location where the specific coil is located. Because the magnetic field can vary as a function of space, as shown in FIG. 12A, the field and frequency of the coil can be adjusted to approximately match the spatial location. Here the coils can be designed to image the field locations Bl, B2, and B3, which are physically separated along a single axis.

[0173] For this low field system, in some cases, a low-noise preamplifier can be designed and configured to leverage the low signal environment of the MRI system. This low noise amplifier can be configured to utilize components that do not generate significant electronic and voltage noise at the desired frequencies (for example, < 3 MHz and >2 MHz). Typical junction fieldWSGR Docket No. 49880-719601effect transistor designs (J-FET) may not have the appropriate noise characteristics at this frequency and can create high frequency instabilities at the GHz range that can bleed into, although several decades of dB lower, into the measured frequency range. Since the gain of the system can preferably be, for example, > 80 dB overall, any small instabilities or intrinsic electrical noise can be amplified and degrade signal integrity.

[0174] Referring to FIG. 12B, RF-RX coils can be designed to image specific limited field of views based upon the target anatomy. The vagina and prostate, for example, extend to about 100 millimeters deep within the human body (see FIG. 12D). Thus, an RX coil for vaginal or prostate imaging can be able to image at least about 100 mm deep inside the human body.According to the Biot Savart law, the magnetic field of a loop coil can be calculated by the following equation,2n * R2* Iwhere pO = 4'7i * 10-7H / m is the vacuum permeability, R is the radius of the loop coil, z is distance along the center line of the coil from its center, and I is the current on the coil (see FIG. 12B). Assuming 1 = 1 Ampere, with the goal of locating a figure of magnetic field (Bz) at z = 100 mm, the maximum position is when R is 140 mm, as shown in FIG. 12C. In some cases, the RF RX coil network is configured for imaging external to the surface by a distance ranging from about 80 mm to about 120 mm to account for anatomical differences across patients.

[0175] Based upon the geometrical constraints of the body, the loop coil can be set up at the space between the human legs upon the torso. These anatomical constraints provide a difficulty for fitting a 280-mm diameter coil at this location. According to FIG. 12C, the Bz field value is proportional to the radius of the loop when R is less than 140mm. As such, it is advantageous that the coil approach a diameter as large as can be accommodated. For example, the largest loop coil that can be placed between a person’s legs can be about 10 cm large.

[0176] As the size of the coil is limited by the space between legs, the magnetic field of a 10-cm diameter coil may not be capable of reaching the depth of the vagina or prostate. Therefore, a single coil may not be enough for POP imaging. Thus, multiple coils can prove beneficial in receiving signal from different directions. In various embodiments of the MRI system, the magnetic field is provided in the z-direction and RF coils are sensitive to x- and y-direction. In this example case, a loop coil in x-y plane may not collect RF signal from a human since it is sensitive to z-direction, while a butterfly coil may be useful. Based on the location and orientation of an RF coil, the RF coil can be a loop coil or a butterfly coil. In addition, an RF coil can be placed under the body of a subject without limiting the size of the RF coil.WSGR Docket No. 49880-719601

[0177] In some embodiments, the MRI system comprises a plurality of RX coils. In some embodiments, one or more RX coils of the plurality of RX coils can be decoupled. Decoupling the one or more RX couples can comprise: geometry decoupling, capacitive or inductive decoupling, or low or high impedance pre-amplifier coupling.

[0178] The MRI system, in some cases, can have a variant magnetic field from the magnet, and its strength can vary linearly along the z direction. The RX coils can be located in different positions in z-direction, and each coil can be tuned to different frequencies, which can depend on the location of the coils in the system.

[0179] In some embodiments, the RX coils can be constructed from conductive traces that can be pre-tuned to a desired frequency and printed, for example, on a disposable substrate. In some embodiments, a clinician can place the RX coil (or a plurality of RX coils configured in an array) upon the body at the region of interest for a given procedure and dispose of the coil afterwards. For example, and in some cases, the RX coils can be surface coils, which can be affixed to, e.g., worn or taped to, a patient's body. For other body parts, e.g., an ankle or a wrist, the surface coil might be a single-loop configuration, figure-8 configuration, or butterfly coil configuration wrapped around the region of interest. For regions that require significant penetration depth, e.g., the torso or knee, the coil might consist of a Helmholtz coil pair. The main restriction to the receive coil is similar to other MRI systems: the coil can be sensitive to a plane that is orthogonal to the main magnetic field, BO, axis. FIG. 14 is a schematic view of figure-8 coils 1402 disposed within the housing of an MRI system proximate to the surface 1404. In some embodiments, the figure-8 coils 1402 cover the entirety of an underside of the surface 1404. In some embodiments, the coils can be disposed on an underside of a commode. In some embodiments, the MRI system described herein can comprise the commode.

[0180] In some cases, the coils might be inductively coupled to another loop that is electrically connected to the receive preamplifier. This design can allow for easier and unobstructed access of the receive coils.

[0181] In some cases, the size of coils can be limited by the structure of the human body. For example, the coils' size can be positioned and configured to fit in the space between human legs when imaging the vagina or prostate.

[0182] In some cases, the number of turns and loops of the RX coil can be adapted to cover the entire space between the legs so that the entire imaging volume can be covered.Programmable Logic Controller

[0183] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a programmable logicWSGR Docket No. 49880-719601controller (PLC). PLCs are industrial digital computers which can be designed to operate reliably in harsh usage environments and conditions. PLCs can be designed to handle these types of conditions and environments, not just in the external housing, but in the internal components and cooling arrangements as well. As such, PLCs can be adapted for the control of manufacturing processes, such as assembly lines, or robotic devices, or any activity that requires high reliability control and ease of programming and process fault diagnosis.

[0184] In some cases, the system can contain a PLC that can control the system in pseudo realtime. This controller can manage the power cycling and enabling of the gradient amplifier system, the radio frequency transmission system, the frequency tuning system, and sends a keep alive signal (e.g., a message sent by one device to another to check that the link between the two is operating, or to prevent the link from being broken) to the system watchdog. The system watchdog can continually look for a strobe signal supplied by the computer system. If the computer threads stall, a strobe is missed that can trigger the watchdog to enter a fault condition. If the watchdog enters a fault condition, the watchdog can be operated to depower the system.

[0185] The PLC can handle low level logic functions on incoming and outgoing signals into system. This system can monitor the subsystem health and control when subsystems needed to be powered or enabled. The PLC can be designed in different ways. One design example includes a PLC with one main motherboard with four expansion boards. Due to the speed of the microcontroller on the PLC, subsystems can be managed in pseudo real-time, while real-time applications can be handled by the computer or spectrometer on the system.

[0186] The PLC can serve many functional responsibilities including, for example, powering on / off the gradient amplifiers (discussed in greater detail herein) and the RF amplifier (discussed in greater detail herein), enabling / disabling the gradient amplifiers and the RF amplifier, setting the digital and analog voltages for the RF coil tuning, and strobing the system watchdog.Robot

[0187] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a robot.

[0188] In some medical procedures, such as a prostate biopsy, it is typical for the patient to endure a lengthy procedure in an uncomfortable prone position, which often includes remaining motionless in one specific body position during the entire procedure. In such long procedures, if a metallic ferromagnetic needle is used for the biopsy with guidance from an MRI system, the needle may experience attraction force from the strong magnets of the MRI system, and thus may cause it to deviate from its path during the length of the procedure. Even in the case of usingWSGR Docket No. 49880-719601a non-magnetic needle, the local field distortions can cause distortions in the magnetic resonance images, and therefore, the image quality surrounding the needle may result in a poor quality. To avoid such distortions, pneumatic robots with complex compressed air mechanism have been designed to work in conjunction with conventional MRI systems. Even then, access to target anatomy remains challenging due to the form factor of currently available MRI systems.

[0189] The various embodiments presented herein include improved MRI systems that are configured to use for guiding in medical procedures, including, for example, robot-assisted, invasive medical procedures. The technologies, methods and apparatuses disclosed herein relate to a guided robotic system using magnetic resonance imaging as a guidance to automatically guide a robot (referred to herein as "a robotic system") in medical procedures. In some cases, the disclosed technologies combine a robotic system with magnetic resonance imaging as guidance. In some cases, the robotic system disclosed herein is combined with other suitable imaging techniques, for example, ultrasound, x-ray, laser, or any other suitable diagnostic or imaging methodologies.Spectrometer

[0190] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a spectrometer.

[0191] A spectrometer can operate to control all real-time signaling used to generate images. It creates the RF transmission (RF-TX) waveform, gradient waveforms, frequency tuning trigger waveform, and blanking bit waveforms. These waveforms are then synchronized with the RF receiver (RF-RX) signals. This system can generate frequency swept RF-TX pulses and phase cycled RF-TX pulses. The swept RF-TX pulses allow for an inhomogeneous B1+ field (RF-TX field) to excite a sample volume more effectively and efficiently. It can also digitize multiple RF RX channels with the current configuration set to four receiver channels. However, this system architecture allows for an easy system scale-up to increase the number of transmit and receive channels to a maximum of 32 transmit channels and 16 receive channels without having to change the underlying hardware or software architecture.

[0192] The spectrometer can serve many functional responsibilities including, for example, generating and synchronizing the RF-TX (discussed in greater detail herein) waveforms, X gradient waveforms, Y-gradient waveforms, blanking bit waveforms, frequency tuning trigger waveform and RF-RX windows, and digitizing and signal processing the RF-RX data using, for example, quadrature demodulation followed by a finite impulse response filter decimation such as, for example, a cascade integrating comb (CIC) filter decimation.WSGR Docket No. 49880-719601

[0193] The spectrometer can be designed in different ways. One design example includes a spectrometer with three main components: 1) a first software design radio (SDR 1) operating with Basic RF-TX daughter cards and Basic RF-RX daughter cards; 2) a second software design radio (SDR 2) operating with LFRF TX daughter cards and Basic RF-RX daughter cards; and 3) a clock distribution module (octoclock) that can synchronize the two devices.

[0194] SDRs are the real-time communication device between the transmitted signals and received MRI signals. They can communicate over 10-Gbit optical fiber to the computer using a Small Form-factor Pluggable Plus transceiver (SFP+) communication protocol. This communication speed can allow the waveforms to be generated with high fidelity and high reliability.

[0195] Each SDR can include a motherboard with an integrated field-programmable gate array (FPGA), digital to analog converters, analog to digital converters, and four module slots for integrating different daughtercards. Each of these daughtercards can function to change the frequency response of the associated TX or RX channel. In some cases, the system can utilize many variations of daughtercards including, for example, a Basic RF version, and a low frequency (LP) RF version. The Basic RF daughtercards can be used for generating and measuring RF signals. The LP RF version can be used for generating gradient, trigger and blanking bit signals.

[0196] The octoclock can be used to synchronize a multi-channel SDR system to a common timing source while providing high-accuracy time and frequency reference distribution. It can do so, for example, with 8-way time and frequency distribution (1 PPS and 10MHz). An example of an octoclock is the Ettus Octoclock CD A, which can distribute a common clock to up to eight SDRs to ensure phase coherency between the two or more SDR sources.RF Amplifier / Gradient Amplifier

[0197] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a radio frequency amplifier (RF amplifier) and a gradient amplifier.

[0198] A RF amplifier is a type of electronic amplifier that can convert a low-power radiofrequency signal into a higher power signal. In operation, the RF amplifier can accept signals at low amplitudes and provide, for example, up to 60 dB of gain with a flat frequency response. This amplifier can accept three phase AC input voltage and can have a 10% max duty cycle. The amplifier can be gated by a 5 V digital signal so that unwanted noise is not generated when the MRI is receiving signal.WSGR Docket No. 49880-719601

[0199] In operation, a gradient amplifier can increase the energy of the signal before it reaches the gradient coils such that the field strength can be intense enough to produce the variations in the main magnetic field for localization of the later received signal. The gradient amplifier can have two active amplification channels that can be controlled independently. Each channel can send out current to either the X or Y channel. The third axis of spatial encoding can be handled by a permanent gradient in the main magnetic field (BO). With varying combinations of pulse sequences, the signal can be localized in three dimensions and reconstructed to create an object.Display / GUI

[0200] As discussed herein, and in some cases, the various systems, and various combinations of features that make up the various system embodiments, can also include a display in the form of, for example, a graphical user interface (GUI). In some cases, the GUI can take any contemplated form necessary to convey the information necessary to run magnetic resonance imaging procedures.

[0201] Further, it can be appreciated that the display may be embodied in any of a number of other forms, such as, for example, a rack-mounted computer, mainframe, supercomputer, server, client, a desktop computer, a laptop computer, a tablet computer, hand-held computing device (e.g., PDA, cell phone, smart phone, palmtop, etc.), cluster grid, netbook, embedded systems, or any other type of special or general purpose display device as may be desirable or appropriate for a given application or environment.

[0202] The GUI is a system of interactive visual components for computer software. A GUI can display objects that convey information and represent actions that can be taken by the user. The objects change color, size, or visibility when the user interacts with them. GUI objects include, for example, icons, cursors, and buttons. These graphical elements are sometimes enhanced with sounds, or visual effects like transparency and drop shadows.

[0203] A user can interact with a GUI using an input device, which can include, for example, alphanumeric and other keys, mouse, a trackball or cursor direction keys for communicating direction information and command selections to a processor and for controlling cursor movement on the display. An input device may also be the display configured with touchscreen input capabilities. This input device can have two degrees of freedom in two axes, a first axis (i.e., x) and a second axis (i.e., y), that allows the device to specify positions in a plane.However, it can be understood that input devices allowing for 3 -dimensional (x, y and z) cursor movement are also contemplated herein.WSGR Docket No. 49880-719601

[0204] In some cases, the touchscreen, or touchscreen monitor, can serve as the primary human interface device that allows a user to interact with the MRI. The screen can have a projected capacitive touch sensitive display with an interactive virtual keyboard. The touchscreen can have several functions including, for example, displaying the graphical user interface (GUI) to the user, relaying user input to the system's computer, and starting or stopping a scan.

[0205] In some cases, GUI views can be screens displayed (Qt widgets) to the user with appropriate buttons, edit fields, labels, images, etc. These screens can be constructed using a designer tool such as, for example, the Qt designer tool, to control placement of widgets, their alignment, fonts, colors, etc. A user interface (UI) sub controller can possess modules configured to control the behavior (display and responses) of the respective view modules.

[0206] Several application utilities (App Util) modules can perform specific functions. For example, S3 modules can handle data communication between the system and, for example, Amazon Web Services (AWS). Event Filters can be present to ensure valid characters are displayed on screen when user inputs are required. Dialog messages can be used to show various status, progress messages, or require user prompts. Moreover, a system controller module can be utilized to handle coordination between the sub controller modules, and key data processing blocks in the system, the pulse sequence generator, pulse interpreter, spectrometer and reconstruction.Pre-Polarizer

[0207] As discussed herein, and in some cases, the various workflows or methods, and various combinations of steps that make up the various workflow or method embodiments, can also include a pre-polarization step.

[0208] In some embodiments, the prepolarizer can be charged by a system power supply. The powering of this polarizer can temporarily change the magnetic field within the field of view either by increasing or decreasing the main magnetic field strength. This change in the magnetic field then creates a change in the total number of nuclear spins that are aligned within the field of view and it changes the time constants by which the nuclear spins relax. An increase in the field allows for more nuclear spins to be aligned with the field, thus temporarily increasing the signal from a given voxel. A decrease in the field changes the relaxation properties of the objects and can allow for increased contrast within the field of view.

[0209] In some cases, the prepolarizer might be first charged to increase the field strength and therefore the signal strength. Then after waiting an appropriate amount of time for the nuclear spins to align (as dictated by the T1 time of the desired spins), the prepolarizer can be removed.WSGR Docket No. 49880-719601As this prepolarizer is depowered, the spins that are aligned will begin to relax and lose energy but can still be imaged by the magnetic resonance system at an increased signal level than when the system did not apply a prepolarizing pulse.Patient Intake

[0210] As discussed herein, and in some cases, the various workflows or methods, and various combinations of steps that make up the various workflow or method embodiments, can also include a patient intake step.

[0211] As part of this step, and any relevant information can be part of the patient intake step, including the intake of all data relevant to the performance of the magnetic resonance system, In some cases herein.

[0212] In some cases, the patient intake step can include, not only data inputted by user, but also data downloaded from any memory source, whether it be, for example, data from a remote data storage component (e.g., the cloud), an on-board data storage component, or portable data storage component (e.g., external flash / solid state drives and external hard drives).

[0213] In some cases, and further related to memory sources, an on-board data storage component (e.g., on board a computing system within an MRI system) can be a random-access memory (RAM) or other dynamic memory, or a read only memory (ROM) or other static storage device.

[0214] In some cases, and further related to memory sources, a remote or portable data storage component can include, for example, a magnetic disk, optical disk, solid state drive (SSD), and a media drive and a removable storage interface. A media drive may include a drive or other mechanism to support fixed or removable storage media, such as a hard disk drive, a floppy disk drive, a magnetic tape drive, an optical disk drive, a CD or DVD drive (R or RW), flash drive, or other removable or fixed media drive. As these examples illustrate, the storage media may include a computer-readable storage medium having stored therein particular computer software, instructions, or data.

[0215] In some cases, a storage device may include other similar instrumentalities for allowing computer programs or other instructions or data to be loaded into computing system. Such instrumentalities may include, for example, a removable storage unit and an interface, such as a program cartridge and cartridge interface, a removable memory (for example, a flash memory or other removable memory module) and memory slot, and other removable storage units and interfaces that allow software and data to be transferred from the storage device to computing system.WSGR Docket No. 49880-719601

[0216] In some cases, the data types that can be user inputted, uploaded, downloaded, etc., can include, for example, patient name, patient sex, patient weight, patient height, patient contact information, patient birthdate, patient's referring physician, and patient race. In addition, a clinical baseline can be user inputted that includes information such as the patient's Gleason score for any past biopsies, the frequency of sexual intercourse, the last time the patient had food, and the patient's PSA level.Patient Positioning

[0217] As discussed herein, and in some cases, the various workflows or methods, and various combinations of steps that make up the various workflow or methods, can also include a patient positioning step. The patient positions described below comprise exemplary patient positions. In some cases, a patient may have multiple scans scheduled, such that the patient may be positioned in one manner for one scan (e.g., such as shown in FIGS. 13A-13X) and in another way (e.g., supine or seated) for a pelvic scan.

[0218] As a precursor to the positioning, a patient can undergo a patient preparation and screening process, whereby the patient is screened for foreign bodies and devices such as pacemakers that may represent a contraindication to imaging. The patient's important health conditions, including allergies, as well as patient data received as part of the patient intake process, can also be reviewed.

[0219] For positioning in a full-body MRI, a patient can be placed on a table, such as in a supine position. Receiver imaging coils can be arranged around the body part of interest (e.g., pelvis, head, chest, knee, etc.) If EKG or respiratory gating is required, then these devices are attached at this time. A key anatomic structure such as the bridge of the nose or umbilicus is identified as a landmark using laser guidance, and this is correlated with table position by pressing a button on the gantry.

[0220] In some cases, using the example system illustrated in FIGS. 13A-13X as a basis herein, a patient is positioned in any number of different positions depending on the type of anatomical scan.

[0221] As illustrated in FIG. 13A, when the abdomen is the region scanned, the patient can be laid on a surface at a lateral position. As illustrated, for the abdominal scan, a patient can be positioned to lay sideways facing the bore, with the arm closest to the table stretched out and the other at the side of the body. The abdomen region can be positioned such that it is directly in front of the bore.WSGR Docket No. 49880-719601

[0222] As illustrated in FIG. 13B, when an appendage (e.g., arm or hand) is the region scanned, the patient can be laid on a surface at a supine position. As illustrated, for the appendage scan, a patient can be positioned to be laid down with the arm or hand to be scanned situated directly in front of the bore.

[0223] As illustrated in FIG. 13C, when an appendage (e.g., arm or hand) is the region scanned, the patient can also be placed at a seated position. As illustrated, for the appendage scan, a patient can be positioned to be seated with arm to be scanned raised up against the system such that it is situated directly in front of the bore.

[0224] As illustrated in FIG. 13D, when an appendage (e.g., elbow) is the region scanned, the patient can also be placed at a seated position. As illustrated, for the appendage scan, a patient can be positioned to be seated with elbow to be scanned raised up against the system such that it is situated directly in front of the bore and the other arm resting comfortably.

[0225] As illustrated in FIG. 13E, when an appendage (e.g., knee) is the region scanned, the patient can also be situated to stand with the one leg lifted that is to be scanned. As illustrated, for the appendage scan, a patient can be positioned to be standing and facing the bore such that he leg of interest is lifted with the knee resting directly in front of the bore and the other leg placed firmly on the ground for stability.

[0226] As illustrated in FIG. 13F, when an appendage (e.g., knee) is the region scanned, the patient can also be situated in a lateral position. As illustrated, for the appendage scan, a patient can be positioned to lay sideways facing the bore, with the leg of interest bent and the other leg resting on the table and extended out. The patient's knee can be placed such that it is directly in front of the bore.

[0227] As illustrated in FIG. 13G, when an appendage (e.g., foot) is the region scanned, the patient can also be situated in a lateral position. As illustrated, for the appendage scan, a patient can be positioned to lay sideways facing away from the bore, with the leg of interest bent and resting on the table and the other leg extended out. The patient's foot can be placed such that it is directly in front of the bore.

[0228] As illustrated in FIG. 13H, when an appendage (e.g., foot) is the region scanned, the patient can also be situated in a seated position. As illustrated, for the appendage scan, a patient can be positioned to be seated facing the bore, with the leg of interest extended out toward the bore and the other leg resting comfortably. The patient's foot can be placed such that it is directly in front of the bore.

[0229] As illustrated in FIG. 131, when an appendage (e.g., wrist) is the region scanned, the patient can be situated in a seated position. As illustrated, for the appendage scan, a patient canWSGR Docket No. 49880-719601be positioned to be seated parallel to the system, such that the wrist of interest is directly in front of the bore with and the other arm is resting comfortably to the side.

[0230] As illustrated in FIG. 13J, when the breast is the region scanned, the patient can be laid on a surface in a lateral position. As illustrated, for the breast scan, a patient can be positioned to lay sideways facing the bore, with one arm extended out above the head and the other hand resting to the side of the body. The breast region can be positioned to be directly in front of the bore.

[0231] As illustrated in FIG. 13K, when the breast is the region scanned, the patient can also be placed at a seated position. As illustrated, for the breast scan, a patient can be positioned to be seated and facing the bore such that arms are extended out and resting on the top of the system. The breast region can be positioned to be directly in front of the bore.

[0232] As illustrated in FIG. 13L, when the breast is the region scanned, the patient can also be placed at a kneeling position. As illustrated, for the breast scan, a patient can be positioned to be kneeling and facing the bore such that arms are extended out and resting on the top of the system. The breast region can be positioned to be directly in front of the bore.

[0233] As illustrated in FIG. 13M, when the head is the region scanned, the patient can be laid on a surface at a lateral position. As illustrated, for the head scan, a patient can be positioned to lay sideways facing away from the bore, with the head placed directly in front of the bore.

[0234] As illustrated in FIG. 13N, when the head is the region scanned, the patient can also be laid on a surface at a supine position. As illustrated, for the head scan, a patient can be positioned to lay down face up, with the top of the head against the system, such that it is situated directly in front of the bore.

[0235] As illustrated in FIG. 130, when the heart is the region scanned, the patient can be placed at a seated or standing position. As illustrated, for the heart scan, a patient can be positioned to be seated facing the bore such that the heart region is situated directly in front of the bore.

[0236] As illustrated in FIG. 13P, when the kidney is the region scanned, the patient can be laid on a surface at a lateral position. As illustrated, for the kidney scan, a patient can be positioned to lay sideways facing the bore, with the arm closest to the table stretched out and the other at the side of the body. The kidney region can be positioned such that it is directly in front of the bore.

[0237] As illustrated in FIG. 13Q, when the liver is the region scanned, the patient can be laid on a surface at a lateral position. As illustrated, for the liver scan, a patient can be positioned to lay sideways facing the bore, with the arm closest to the table stretched out or bent to rest theWSGR Docket No. 49880-719601head, and the other at the side of the body. The liver region can be positioned such that it is directly in front of the bore.

[0238] As illustrated in FIG. 13R, when the lung is the region scanned, the patient can be placed at a seated position. As illustrated, for the lung scan, a patient can be positioned to be seated facing away from the bore such that the lung region is situated directly in front of the bore.

[0239] As illustrated in FIG. 13S, when the neck is the region scanned, the patient can be laid on a surface at a lateral position. As illustrated, for the neck scan, a patient can be positioned to lay sideways and face away from the bore. The neck region can be positioned to be directly in front of the bore.

[0240] As illustrated in FIG. 13T, when the pelvis is the region scanned, the patient can be laid on a surface at a lithotomy position. As illustrated, for the pelvic scan, a patient can be positioned to have their back resting on the table and legs raised up to be resting against the top of the system. The pelvic region can be positioned to be directly in front of the bore.

[0241] As illustrated in FIG. 13U, when the pelvis is the region scanned, the patient can also be laid on a surface at a lateral position. As illustrated, for the pelvic scan, a patient can be positioned to lay sideways and face away from the bore. The pelvic region of the body can be positioned to be directly in front of the bore.

[0242] As illustrated in FIG. 13V, when the pelvis is the region scanned, the patient can also be placed at a prone position. As illustrated, for the pelvic scan, a patient can be positioned to rest with the chest against a surface, facing away from the bore. The pelvic region can be positioned such that it is directly in front of the bore.

[0243] As illustrated in FIG. 13W, when the shoulder is the region scanned, the patient can be placed at a seated position. As illustrated, for the shoulder scan, a patient can be positioned to be seated next to the system with the shoulder to be scanned situated directly in front of the bore.

[0244] As illustrated in FIG. 13X, when the spine is the region scanned, the patient can be placed at a seated position. As illustrated, for the spine scan, a patient can be positioned to be seated with back facing away from the bore and spine situated directly in view of the bore.Biopsy Guidance

[0245] As discussed herein, and in some cases, the various workflows or methods, and various combinations of steps that make up the various workflow or method embodiments, can also include biopsy guidance using the disclosed MRI system.

[0246] In some cases, the procedure for biopsy guidance using the disclosed MRI system may include one from the list of medical procedures consisting of transperineal biopsy, transperinealWSGR Docket No. 49880-719601LDR brachytherapy, transperineal HDR brachytherapy, transperineal laser ablation, transperineal cryoablation, transrectal HIFU, breast biopsies, deep brain stimulation (DBS), brain biopsy, liver biopsy, kidney biopsy, lung biopsy, coronary stent insertion, brain stent insertion, and intensity modulated radiation treatment guidance.Calibration

[0247] As discussed herein, and in some cases, the various workflows or methods, and various combinations of steps that make up the various workflow or method embodiments, can also include a calibration step.

[0248] Calibration can take many forms of processes. In some cases, calibration involves running a full scan, similar to the scan run on a patient, in order to ensure image quality. In some cases, after a predetermined period, a user can be prompted to initiate a calibration routine such as, for example, a RF calibration routine. As part of initiating a calibration, a calibration phantom is positioned to allow calibration to advance. A calibration phantom can take many forms. In some cases, a calibration phantom can be an object (such as an artificial object) of known size and composition that is imaged to test, adjust or monitor an MRI systems homogeneity, imaging performance and orientation aspects. A phantom can be a fluid filled container or bottle often filled with a plastic structure of various sizes and shapes.

[0249] RF Calibration routine, in particular, optimizes RF pulse parameters such as, for example, signal power, signal duration and signal bandwidth to ensure image quality. The calibration routine acquires signal data from a calibration phantom using a predetermined set of parameters and sequence. Calibration data can be processed to determine the parameter set that can be used during imaging scans.Processing Module

[0250] As discussed herein, and in some cases, the various workflows or methods, and various combinations of steps that make up the various workflow or method embodiments, can also include a processing module.

[0251] In some cases, a processing module serves many functions. For example, a processing module can operate to receive signal data acquired during the scan, process the data, and reconstruct those signals to produce an image that can be viewed (for example, via a touchscreen monitor that displays a GUI to the user), analyzed and annotated by system users. To create an image, an NMR signal can be localized in three-dimensional space. Magnetic gradient coils localize the signal and are operated before or during the RF acquisition. By prescribing a RF andWSGR Docket No. 49880-719601gradient coil application sequence, called a pulse sequence, the signals acquired correspond to a specific magnetic field and RF field arrangement. Using mathematical operators and image reconstruction techniques, arrays of these acquired signals can be reconstructed into an image. These images can be generated from simple linear combinations of magnetic field gradients. In some cases, the system can operate to reconstruct the acquired signals from a-priori knowledge of, for example, the gradient fields, RF fields, and pulse sequences.

[0252] In some cases, the processing module can also operate to compensate for patient motion during a scan procedure. Motion (e.g., beating heart, breathing lungs, bulk patient movement) is one of the most common sources of artifacts in MRI, with such artifacts affecting image quality by leading to misinterpretations in the images and a subsequent loss in diagnostic quality.Therefore, motion compensation protocols can help address these issues at minimal cost in time, spatial resolution, temporal resolution, and signal-to-noise ratio.

[0253] In some cases, the processing module might include artificial intelligence machine learning modules designed to denoise the signal and improve the image signal-to-noise ratio.

[0254] In some cases, the processing module can also operate to assist clinicians in planning a path for subsequent patient intervention procedures, such as biopsy. In some cases, a robot can be provided as part of the system to perform the intervention procedure. The processing module can communicate instructions to the robot, based on image analysis, to properly access, for example, the appropriate region of the body requiring a biopsy.

[0255] In some cases, the radio frequency transmit coil and the single-sided gradient coil set of the MRI can be located on the front surface. In some cases, the front surface is a concave surface. In some cases, the permanent magnet of the MRI has an aperture through the center of the permanent magnet. In some cases, the static magnetic field of the permanent magnet ranges from 1 mT to 1 T. In some cases, the static magnetic field of the permanent magnet ranges from 10 mT to 195 mT.

[0256] In some cases, the radio frequency transmit coil includes a first ring and a second ring that are connected via one or more capacitors and / or one or more rungs. In some cases, the radio frequency transmit coil is non-planar and oriented to partially surround the region of interest. In some cases, the single-sided gradient coil set is non-planar and oriented to partially surround the region of interest. In some cases, the single-sided gradient coil set is configured to project a magnetic field gradient to the region of interest. In some cases, the single-sided gradient coil set includes one or more first spiral coils at a first position and one or more second spiral coils at a second position, the first position and the second position being located opposite each otherWSGR Docket No. 49880-719601about a center region of the single-sided gradient coil set. In some cases, the single-sided gradient coil set has a rise time less than 10 ps.

[0257] In some cases, the electromagnet is configured to alter the static magnetic field of the permanent magnet within the region of interest. In some cases, the electromagnet has a magnetic field strength from 10 mT to 1 T. In some cases, the radio frequency receive coil is a flexible coil configured to be affixed to an anatomical portion of a patient for imaging within the region of interest. In some cases, the radio frequency receive coil is in one of a single-loop coil configuration, figure-8 coil configuration, or butterfly coil configuration, wherein the coil is smaller than the region of interest. In some cases, the radio frequency transmit coil and the single-sided gradient coil set are concentric about the region of interest. In some cases, the magnetic resonance imaging system is a single-sided magnetic resonance imaging system that comprises a bore having an opening positioned about a center region of the front surface.

[0258] FIG. 15 is a flowchart for a method S300 of performing a scan on a magnetic resonance imaging system, according to various embodiments. In some cases, the method S300 includes at step S310 providing a housing having a front surface, a permanent magnet for providing a static magnetic field, a radio frequency transmit coil, and a single-sided gradient coil set. In some cases, the radio frequency transmit coil and the single-sided gradient coil set are positioned proximate to the front surface. In some cases, the method S300 includes providing an electromagnet at step S320. In some cases, the method S300 includes at step S330 activating at least one of the radio frequency transmit coil, the single-sided gradient coil set, or the electromagnet to generate an electromagnetic field in a region of interest. In some cases, the region of interest resides outside the front surface.

[0259] In some cases, the method S300 includes activating a radio frequency receive coil to obtain imaging data at step S340, reconstructing obtained imaging data to produce an output image for analysis at step S350 and displaying the output image for user review and annotation at step S360.

[0260] In some cases, the radio frequency transmit coil and the single-sided gradient coil set are located on the front surface. In some cases, the front surface is a concave surface. In some cases, the permanent magnet has an aperture through center of the permanent magnet. In some cases, the static magnetic field of the permanent magnet ranges from 1 mT to 1 T. In some cases, the static magnetic field of the permanent magnet ranges from 10 mT to 195 mT.

[0261] In some cases, the radio frequency transmit coil includes a first ring and a second ring that are connected via one or more capacitors and / or one or more rungs. In some cases, the radio frequency transmit coil is non-planar and oriented to partially surround the region of interest. InWSGR Docket No. 49880-719601some cases, the single-sided gradient coil set is non-planar and oriented to partially surround the region of interest. In some cases, the single-sided gradient coil set is configured to project a magnetic field gradient to the region of interest. In some cases, the single-sided gradient coil set includes one or more first spiral coils at a first position and one or more second spiral coils at a second position, the first position and the second position being located opposite each other about a center region of the single-sided gradient coil set. In some cases, the single-sided gradient coil set has a rise time less than 10 ps.

[0262] In some cases, the electromagnet is configured to alter the static magnetic field of the permanent magnet within the region of interest. In some cases, the electromagnet has a magnetic field strength from 10 mT to 1 T. In some cases, the radio frequency receive coil is a flexible coil configured to be affixed to an anatomical portion of a patient for imaging within the region of interest. In some cases, the radio frequency receive coil is in one of a single-loop coil configuration, figure-8 coil configuration, or butterfly coil configuration, wherein the coil is smaller than the region of interest. In some cases, the radio frequency transmit coil and the single-sided gradient coil set are concentric about the region of interest. In some cases, the magnetic resonance imaging system is a single-sided magnetic resonance imaging system that comprises a bore having an opening positioned about a center region of the front surface.

[0263] FIG. 16 is a flowchart for a method S400 of performing a scan on a magnetic resonance imaging system, according to various embodiments. In some cases, the method S400 includes at step S410 providing a housing having a concave front surface, a permanent magnet for providing a static magnetic field, a radio frequency transmit coil, and a single-sided gradient coil set. In some cases, the radio frequency transmit coil and the single-sided gradient coil set are positioned proximate to the front surface.

[0264] In some cases, the method S400 includes at step S420 activating at least one of the radio frequency transmit coil and the at least one gradient coil set to generate an electromagnetic field in a region of interest. In some cases, the region of interest resides outside the concave front surface.

[0265] In some cases, the method S400 includes activating a radio frequency receive coil to obtain imaging data at step S430, reconstructing obtained imaging data to produce an output image for analysis at step S440 and displaying the output image for user review and annotation at step S450.

[0266] In some cases, the radio frequency transmit coil and the single-sided gradient coil set are located on the concave front surface. In some cases, the static magnetic field of the permanent magnet ranges from 1 mT to IT. In some cases, the static magnetic field of the permanentWSGR Docket No. 49880-719601magnet ranges from 10 mT to 195 mT. In some cases, the radio frequency transmit coil comprises a first ring and a second ring that are connected via one or more capacitors and / or one or more rungs. In some cases, the radio frequency transmit coil is non-planar and oriented to partially surround the region of interest. In some cases, the at least one gradient coil set is non-planar, single sided, and oriented to partially surround the region of interest. In some cases, the at least one gradient coil set is configured to project magnetic field gradient in the region of interest.

[0267] In some cases, the at least one gradient coil set comprises one or more first spiral coils at a first position and one or more second spiral coils at a second position, the first position and the second position being located opposite each other about a center region of the at least one gradient coil set. In some cases, the at least one gradient coil set has a rise time less than 10 ps. In some cases, the permanent magnet has an aperture through the center of the permanent magnet. In some cases, the system further includes an electromagnet configured to alter the static magnetic field of the permanent magnet within the region of interest. In some cases, the electromagnet has a magnetic field strength from 10 mT to 1 T. In some cases, the radio frequency receive coil is a flexible coil configured to be affixed to an anatomical portion of a patient for imaging within the region of interest. In some cases, the radio frequency receive coil is in one of a single-loop coil configuration, figure-8 coil configuration, or butterfly coil configuration, where the coil is smaller than the region of interest.

[0268] In some cases, the radio frequency transmit coil and the at least one gradient coil set are concentric about the region of interest. In some cases, the magnetic resonance imaging system is a single-sided magnetic resonance imaging system that comprises a magnetic resonance imaging scanner or a magnetic resonance imaging spectrometer.Computing Systems

[0269] FIG. 17 shows the computer system 1701 comprising a central processing unit (CPU, also “processor” and “computer processor” herein) 1705, which can be a single core or multi core processor, or a plurality of processors for parallel processing. The computer system 1701 also includes memory or memory location (e.g., random-access memory, read-only memory, flash memory), electronic storage unit 1715 (e.g., hard disk), communication interface 1720 (e.g., network adapter) for communicating with one or more other systems, and peripheral devices 1725, such as cache, other memory, data storage and / or electronic display adapters. The memory, storage unit 1715, interface 1720 and peripheral devices 1725 are in communication with the CPU 1705 through a communication bus (solid lines), such as a motherboard. TheWSGR Docket No. 49880-719601storage unit 1715 can be a data storage unit (or data repository) for storing data. The computer system 1701 can be operatively coupled to a computer network (“network”) 1730 with the aid of the communication interface 1720. The network 1730 can be the Internet, an internet and / or extranet, or an intranet and / or extranet that is in communication with the Internet. The network 1730 in some cases is a telecommunication and / or data network. The network 1730 can include one or more computer servers, which can enable distributed computing, such as cloud computing. The network 1730, in some cases with the aid of the computer system 1701, can implement a peer-to-peer network, which may enable devices coupled to the computer system 1701 to behave as a client or a server.

[0270] The CPU 1705 can execute a sequence of machine-readable instructions, which can be embodied in a program or software. The instructions may be stored in a memory location, such as the memory. The instructions can be directed to the CPU 1705, which can subsequently program or otherwise configure the CPU 1705 to implement methods of the present disclosure. Examples of operations performed by the CPU 1705 can include fetch, decode, execute, and writeback.

[0271] The CPU 1705 can be part of a circuit, such as an integrated circuit. One or more other components of the system 1701 can be included in the circuit. In some cases, the circuit is an application specific integrated circuit (ASIC).

[0272] The storage unit 1715 can store files, such as drivers, libraries and saved programs. The storage unit 1715 can store user data, e.g., user preferences and user programs. The computer system 1701 in some cases can include one or more additional data storage units that are external to the computer system 1701, such as located on a remote server that is in communication with the computer system 1701 through an intranet or the Internet.

[0273] The computer system 1701 can communicate with one or more remote computer systems through the network 1730. For instance, the computer system 1701 can communicate with a remote computer system of a user. Examples of remote computer systems include personal computers (e.g., portable PC), slate or tablet PC’s (e.g., Apple® iPad, Samsung® Galaxy Tab), telephones, Smart phones (e.g., Apple® iPhone, Android-enabled device, Blackberry®), or personal digital assistants. The user can access the computer system 1701 via the network 1730.

[0274] Methods as described herein can be implemented by way of machine (e.g., computer processor) executable code stored on an electronic storage location of the computer system 1701, such as, for example, on the memory 1710 or electronic storage unit 1715. The machine executable or machine-readable code can be provided in the form of software. During use, the code can be executed by the processor 1705. In some cases, the code can be retrieved from theWSGR Docket No. 49880-719601storage unit 1715 and stored on the memory 1710 for ready access by the processor 1705. In some situations, the electronic storage unit 1715 can be precluded, and machine-executable instructions are stored on memory 1710.

[0275] The code can be pre-compiled and configured for use with a machine having a processer adapted to execute the code or can be compiled during runtime. The code can be supplied in a programming language that can be selected to enable the code to execute in a pre-compiled or as-compiled fashion.

[0276] Aspects of the systems and methods provided herein, such as the computer system 1701, can be embodied in programming. Various aspects of the technology may be thought of as “products” or “articles of manufacture” typically in the form of machine (or processor) executable code and / or associated data that is carried on or embodied in a type of machine-readable medium. Machine-executable code can be stored on an electronic storage unit, such as memory (e.g., read-only memory, random-access memory, flash memory) or a hard disk.“Storage” type media can include any or all of the tangible memory of the computers, processors or the like, or associated modules thereof, such as various semiconductor memories, tape drives, disk drives and the like, which may provide non-transitory storage at any time for the software programming. All or portions of the software may at times be communicated through the Internet or various other telecommunication networks. Such communications, for example, may enable loading of the software from one computer or processor into another, for example, from a management server or host computer into the computer platform of an application server. Thus, another type of media that may bear the software elements includes optical, electrical and electromagnetic waves, such as used across physical interfaces between local devices, through wired and optical landline networks and over various air-links. The physical elements that carry such waves, such as wired or wireless links, optical links or the like, also may be considered as media bearing the software. As used herein, unless restricted to non-transitory, tangible “storage” media, terms such as computer or machine “readable medium” refer to any medium that participates in providing instructions to a processor for execution.

[0277] Hence, a machine-readable medium, such as computer-executable code, may take many forms, including but not limited to, a tangible storage medium, a carrier wave medium or physical transmission medium. Non-volatile storage media include, for example, optical or magnetic disks, such as any of the storage devices in any computer(s) or the like, such as may be used to implement the databases, etc. shown in the drawings. Volatile storage media include dynamic memory, such as main memory of such a computer platform. Tangible transmission media include coaxial cables; copper wire and fiber optics, including the wires that comprise aWSGR Docket No. 49880-719601bus within a computer system. Carrier-wave transmission media may take the form of electric or electromagnetic signals, or acoustic or light waves such as those generated during radio frequency (RF) and infrared (IR) data communications. Common forms of computer-readable media therefore include for example: a floppy disk, a flexible disk, hard disk, magnetic tape, any other magnetic medium, a CD-ROM, DVD or DVD-ROM, any other optical medium, punch cards paper tape, any other physical storage medium with patterns of holes, a RAM, a ROM, a PROM and EPROM, a FLASH-EPROM, any other memory chip or cartridge, a carrier wave transporting data or instructions, cables or links transporting such a carrier wave, or any other medium from which a computer may read programming code and / or data. Many of these forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to a processor for execution.

[0278] The computer system 1701 can include or be in communication with an electronic display 1735 that comprises a user interface (UI) 1740 for providing, for example, a UI on a display of the user device. Examples of UI’s include, without limitation, a graphical user interface (GUI) and web-based user interface. In some embodiments, the electronic display may comprise a touch screen.

[0279] Methods and systems of the present disclosure can be implemented by way of one or more algorithms. An algorithm can be implemented by way of software upon execution by the central processing unit 1705. The algorithm can, for example, analyze data obtained by the user identification device or the monitoring module (or the medication monitoring module).Definitions

[0280] Unless defined otherwise, all terms of art, notations and other technical and scientific terms or terminology used herein are intended to have the same meaning as is commonly understood by one of ordinary skill in the art to which the claimed subject matter pertains. In some cases, terms with commonly understood meanings are defined herein for clarity and / or for ready reference, and the inclusion of such definitions herein should not necessarily be construed to represent a substantial difference over what is generally understood in the art.

[0281] Throughout this application, various embodiments may be presented in a range format. It should be understood that the description in range format is merely for convenience and brevity and should not be construed as an inflexible limitation on the scope of the disclosure.Accordingly, the description of a range should be considered to have specifically disclosed all the possible subranges as well as individual numerical values within that range. For example, description of a range such as from 1 to 6 should be considered to have specifically disclosedWSGR Docket No. 49880-719601subranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6 etc., as well as individual numbers within that range, for example, 1, 2, 3, 4, 5, and 6. This applies regardless of the breadth of the range.

[0282] The ranges disclosed herein also encompass any and all overlap, sub-ranges, and combinations thereof. Language such as “up to,” “at least,” “greater than,” “less than,” “between,” and the like includes the number recited. Numbers preceded by a term such as “approximately”, “about”, and “substantially” as used herein include the recited numbers, and also represent an amount close to the stated amount that still performs a desired function or achieves a desired result. The term “about” or “approximately” may mean within an acceptable error range for the particular value, which will depend in part on how the value is measured or determined, e.g., the limitations of the measurement system. For example, the terms “approximately”, “about”, and “substantially” may refer to an amount that is within less than 10% of, within less than 5% of, within less than 1% of, within less than 0.1% of, and within less than 0.01% of the stated amount. For example, “about” may mean within 1 or more than 1 standard deviation, per the practice in the art. Alternatively, “about” may mean a range of up to 20%, up to 10%, up to 5%, or up to 1% of a given value. As used herein, the term “about” a number refers to that number plus or minus 10% of that number. The term “about” a range refers to that range minus 10% of its lowest value and plus 10% of its greatest value. Where particular values are described in the application and claims, unless otherwise stated the term “about” meaning within an acceptable error range for the particular value may be assumed.

[0283] As used in the specification and claims, the singular forms “a”, “an” and “the” include plural references unless the context clearly dictates otherwise. For example, the term “a sample” includes a plurality of samples, including mixtures thereof.

[0284] The terms “determining,” “measuring,” “evaluating,” “assessing,” “assaying,” and “analyzing” are often used interchangeably herein to refer to forms of measurement. The terms include determining if an element is present or not (for example, detection). These terms can include quantitative, qualitative or quantitative and qualitative determinations. Assessing can be relative or absolute. “Detecting the presence of’ can include determining the amount of something present in addition to determining whether it is present or absent depending on the context.

[0285] The terms “subject,” “individual,” or “patient” are often used interchangeably herein. A “subject” can be a biological entity containing expressed genetic materials. The subject can be a mammal. The mammal can be a human. The subject may be diagnosed or suspected of being atWSGR Docket No. 49880-719601high risk for a disease. In some cases, the subject is not necessarily diagnosed or suspected of being at high risk for the disease.

[0286] As used herein, the terms “treatment” or “treating” are used in reference to a pharmaceutical or other intervention regimen for obtaining beneficial or desired results in the recipient. Beneficial or desired results include but are not limited to a therapeutic benefit and / or a prophylactic benefit. A therapeutic benefit may refer to eradication or amelioration of symptoms or of an underlying disorder being treated. Also, a therapeutic benefit can be achieved with the eradication or amelioration of one or more of the physiological symptoms associated with the underlying disorder such that an improvement is observed in the subject, notwithstanding that the subject may still be afflicted with the underlying disorder. A prophylactic effect includes delaying, preventing, or eliminating the appearance of a disease or condition, delaying or eliminating the onset of symptoms of a disease or condition, slowing, halting, or reversing the progression of a disease or condition, or any combination thereof. For prophylactic benefit, a subject at risk of developing a particular disease, or to a subject reporting one or more of the physiological symptoms of a disease may undergo treatment, even though a diagnosis of this disease may not have been made.

[0287] The section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.Examples

[0288] Example 1 - Pre-Processing Subspace Extraction Method in Phantom

[0289] To demonstrate the efficacy of the pre-processing subspace extraction method (Embodiment 1), an experimental validation was performed using both a tissue-mimicking gel block phantom and an American College of Radiology (ACR) approved extremity phantom.

[0290] The experiments were conducted on a low-field MR scanner (approximately 65 mT) operating in an unshielded environment. Primary MR signal data was acquired using a 5-channel receive (Rx) array. Auxiliary EMI data was simultaneously acquired using an EMI detector box comprising three orthogonal probes (X, Y, and Z orientations), as shown in FIG. 5. All probes were tuned to 2.64 MHz to maximize sensitivity to environmental interference.

[0291] During the acquisition, various EMI types were evaluated, including ambient environmental noise and interference artificially introduced using a signal generator at both on-resonance and 10 kHz off-resonance frequencies. The system was tested across multiple pulse sequence configurations, the parameters of which are summarized in Table 1 below.WSGR Docket No. 49880-719601

[0292] Table 1: EMI Test ParametersParameters 2D radial xSPEN 3D T2 Cartesian 3D T2 RadialTE (msec) 5.4 5.3 5.2TR (sec) 1.6 1.8 1.45# or spokes / PE lines 600 128 250# of echoes 12 8 12Acquisition time (min) 96 11.5 6.04FOV (cm x cm x cm) 18 x 18 x 11 18 x 18 x 11 18 x 18 x 10

[0293] FIG. 18 illustrates the spectral performance of the method with three EMI sensors arrayed as shown in FIG. 5. In a control environment, the signal amplitude across the frequency spectrum provides a baseline for interpretation. When inducing 10 kHz off-resonance EMI, the pre-correction signal exhibits a significant noise spike. After applying the subspace extraction method, the post-correction signal demonstrates a Noise Reduction Ratio (NRR) of 2.71 dB, computed as:signalcorrected\NRR = 201og10signalcontroi'

[0294] While preferred embodiments of the present disclosure have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions can occur to those skilled in the art without departing from the present disclosure. It should be understood that various alternatives to the embodiments of the present disclosure described herein may be employed in practicing the present disclosure. It is intended that the following claims define the scope of the invention and that methods and structures within the scope of these claims and their equivalents be covered thereby.

Claims

WSGR Docket No. 49880-719601CLAIMS WHAT IS CLAIMED IS:

1. A method for mitigating electromagnetic interference (EMI) in magnetic resonance (MR) imaging, the method comprising:a. obtaining EMI data from at least one EMI detector;b. obtaining MR signal data associated with an object within an MR imaging region; c. building a noise convolution matrix from the obtained EMI data;d. extracting a noise subspace from the noise convolution matrix by computing a low-rank approximation, thereby generating a truncated basis set representative of dominant EMI components;e. projecting the noise convolution matrix onto the truncated basis set to generate a reduced-dimensionality noise representation;f. computing a dynamic transformation individually for each repetition time (TR) interval to map the reduced-dimensionality noise representation to the obtained MR signal data, thereby generating a noise estimate; andg. subtracting the noise estimate from the obtained MR signal data to generate corrected MR signal data.

2. The method of claim 1, wherein the at least one EMI detector comprises one or more of: an E-field probe, a surface coil, an electrode connected to the patient's skin, a direct connection to a sub-system ground, a direct connection to a patient table, or a 50 Ohm terminated channel.

3. The method of claim 1 or 2, wherein the at least one EMI detector is positioned external to the MR imaging volume.

4. The method of any one of claims 1 to 3, wherein the obtained EMI data comprises data collected using one or more primary MRI receive (Rx) coil elements during a known noise region of an MR pulse sequence.

5. The method of any one of claims 1 to 4, wherein building the noise convolution matrix comprises constructing a multi-dimensional block Hankel convolution matrix using a sliding window operation.

6. The method of any one of claims 1 to 5, wherein the EMI data is obtained from an MR pulse sequence during a predetermined noise-only acquisition region.WSGR Docket No. 49880-7196017. The method of any one of claims 1 to 6, wherein computing the low-rank approximation comprises applying a dimensionality reduction technique comprising one or more of: singular value decomposition (SVD), principal component analysis (PCA), eigenvalue decomposition, or randomized low-rank matrix approximation.

8. The method of claim 7, wherein generating the truncated basis set comprises thresholding singular values or eigenvalues derived from the dimensionality reduction technique based on a predefined fractional threshold.

9. The method of claim 7 or 8, wherein computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors and Rx coil elements utilizing data acquired from noise-only periods.

10. The method of any one of claims 7 to 9, wherein computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors utilizing only data acquired during signal portions of MRI data acquisitions.

11. The method of claim 9 or 10, wherein the noise whitening procedure is applied prior to computing the low-rank approximation using singular value decomposition (SVD).

12. The method of any one of claims 1 to 11, wherein the at least one EMI detector comprises noise coils tuned to frequencies associated with excited bands.

13. The method of claim 12, wherein the noise coils are configured to have high sensitivity to selected frequencies, thereby comprising a high Q-factor.

14. The method of claim 12 or 13, wherein the noise coils are tuned over a broader frequency band and comprise a sensitivity versus frequency profile that matches a frequency distribution of an EMI spectrum over the broader frequency band.

15. The method of any one of claims 1 to 14, wherein a noise correlation matrix and subsequent low-rank approximation are re-computed for specified frequency bands having bandwidths equal to a bandwidth of excited slabs, portions of an excited bandwidth corresponding to slices in a z-dimension, or portions dominated by a particular EMI noise source.WSGR Docket No. 49880-71960116. The method of claim 15, wherein the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only time domain data acquired in a non-excited data acquisition.

17. The method of claim 15, wherein the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only portions of time domain data acquired during an MRI data acquisition.

18. The method of claim 16 or 17, wherein the noise correlation matrix is determined using a sliding window FT on the noise-only time domain data; wherein a sliding window time length is determined by a bandwidth of dominant noise sources or a time length required to obtain statistically significant noise correlation information; and wherein the sliding window FT employs zero filling on negative and positive sides of a time window to improve frequency resolution.

19. The method of any one of claims 1 to 18, wherein a noise correlation computation is applied to data acquired during signal portions of an MR pulse sequence by employing a signal subtraction method to remove MR signal from a signal echo, thereby leaving noise-only data to perform the noise correlation computation using both MRI Rx coil elements and the at least one EMI detector.

20. The method of any one of claims 1 to 19, wherein computing the dynamic transformation comprises fitting the reduced-dimensionality noise representation to the obtained MR signal data using a least-squares estimation to determine transformation coefficients for each individual TR interval.

21. The method of any one of claims 1 to 20, further comprising reconstructing an EMI-corrected MR image using the corrected MR signal data.

22. The method of any one of claims 1 to 21, wherein computing the noise estimate and subtracting the noise estimate are performed jointly within an iterative image reconstruction process.

23. The method of claim 22, wherein the iterative image reconstruction process comprises minimizing a joint objective function that simultaneously estimates a reconstructed MR image and the noise estimate.WSGR Docket No. 49880-71960124. The method of claim 23, wherein the obtained MR signal data comprises k-space data acquired using either a fully sampled acquisition trajectory or an undersampled acquisition trajectory.

25. The method of claim 23 or 24, wherein the joint objective function comprises: a. a data fidelity term that evaluates a difference between the obtained MR signal data and a predicted MR signal derived from the reconstructed MR image and the noise estimate;b. an image regularization term configured to enforce structural properties, on the reconstructed MR image; andc. a noise regularization term configured to enforce a low-rank constraint on the extracted noise subspace.

26. The method of claim 25, wherein the structural properties comprise sparsity or smoothness.

27. The method of claim 25 or 26, wherein the noise regularization term comprises a nuclear norm penalty applied to the noise convolution matrix.

28. The method of any one of claims 25 to 27, wherein minimizing the joint objective function comprises executing an alternating minimization algorithm that alternates between updating the reconstructed MR image and updating the noise estimate.

29. The method of claim 28, wherein updating the noise estimate comprises applying singular value thresholding (SVT) to a data residual to dynamically update the truncated basis set at each iteration of the alternating minimization algorithm.

30. The method of any one of claims 1 to, further comprising:a. performing a pre-scan calibration to identify one or more dominant noise frequency bands within the MR imaging environment; andb. adaptively tuning the at least one EMI detector to the identified one or more dominant noise frequency bands prior to obtaining the EMI data.

31. The method of claim 30, wherein adaptively tuning the at least one EMI detector comprises applying a hardware or software-based bandpass filter configured to maximize sensitivity to the identified one or more dominant noise frequency bands.

32. The method of any one of claims 1 to 31, wherein the truncated basis set is derived at least in part from a historical noise subspace model, wherein the historical noiseWSGR Docket No. 49880-719601subspace model is constructed from accumulated EMI data obtained during a plurality of prior MR imaging sessions at a specific installation site.

33. The method of claim 32, wherein extracting the noise subspace is performed prior to obtaining the MR signal data by computing the low-rank approximation on the accumulated EMI data, thereby generating a pre-computed site-specific basis set configured to bypass realtime basis set generation during a current MR imaging session.

34. The method of claim 32 or 33, wherein computing the noise estimate comprises a two-pass estimation process, comprising:a. a first pass that projects the noise convolution matrix onto the historical noise subspace model to remove baseline environmental interference; andb. a second pass that projects a residual noise representation onto a scan-specific basis set derived from the EMI data obtained during the current MR imaging session to remove transient interference.

35. The method of any one of claims 1 to 34, further comprising:a. comparing a rank or a singular value distribution of the extracted noise subspace to a historical baseline noise subspace model;b. determining a deviation metric between the extracted noise subspace and the historical baseline noise subspace model; andc. triggering an automated system action if the deviation metric exceeds a predetermined threshold.

36. The method of claim 35, wherein the automated system action comprises one or more of generating a user alert indicating elevated environmental interference, dynamically increasing a number of signal averages (NEX) for the MR pulse sequence, or switching to an alternative noise cancellation protocol.

37. The method of any one of claims 1 to 36, wherein extracting the noise subspace comprises processing the noise convolution matrix through a trained artificial neural network comprising an undercomplete autoencoder architecture, wherein a bottleneck layer of the undercomplete autoencoder is configured to output a latent representation that defines the truncated basis set.

38. The method of claim 22, wherein the iterative image reconstruction process is executed by a trained unrolled neural network, wherein individual layers or blocks of the trainedWSGR Docket No. 49880-719601unrolled neural network correspond to iterations of an alternating minimization algorithm, and wherein the trained unrolled neural network comprises learnable parameters configured to dynamically apply singular value thresholding (SVT) to extract the noise subspace.

39. The method of claim 38, wherein the trained unrolled neural network is trained using a supervised learning protocol comprising:a. a training dataset comprising pairs of EMI-corrupted MR k-space data and corresponding ground-truth unaliased MR images; andb. a loss function configured to minimize a difference between an output of the trained unrolled neural network and the ground-truth unaliased MR images.

40. The method of any one of claims 1 to 39, wherein the MR signal data is acquired using a low-field or ultra-1 ow-fi eld MR scanner operating in an unshielded environment lacking a stationary Faraday cage.

41. The method of any one of claims 1 to 40, wherein the MR signal data is acquired using a single-sided MR scanner.

42. The method of any one of claims 1 to 41, wherein the MR signal data is acquired using a portable MR scanner configured for point-of-care operation.

43. The method of any one of claims 1 to 42, wherein the MR signal data is acquired using a pulse sequence comprising non-linear spatiotemporal spatial encoding trajectories.

44. The method of claim 19, wherein the MRI signal used for subtraction is estimated using signals received by a plurality of receive coils and spatial sensitivity relationships between the receive coils.

45. The method of claim 44, wherein estimating the MRI signal includes reconstructing an MRI image and using coil sensitivity profiles to estimate signal contributions in each receive channel.

46. The method of claim 45, wherein the estimated signal contributions are subtracted from the received signals to generate improved estimates of electromagnetic interference present in each receive channel, and wherein the steps of signal estimation, signal subtraction, and noise estimation are repeated iteratively to refine estimates of MRI signal and electromagnetic interference.

47. The method of claim 46, wherein coil sensitivity profiles are estimated or refined using the reconstructed MRI signal and residual signals obtained after signal subtraction.

48. The method of claim 46, wherein spatial signatures of electromagnetic interference across the receive coils are estimated and used to suppress interference components.WSGR Docket No. 49880-71960149. The method of any one of claims 1 to 48, wherein obtaining the MR signal data comprises using a magnetic resonance imaging (MRI) system comprising:a. a housing comprising a surface for contact with a subject; andb. a radio frequency receive (RF RX) coil network;wherein the RF RX coil network is configured to enable imaging in a region of interest, wherein the region of interest is external to the surface of the housing by a distance ranging from about 80 mm to about 120 mm.

50. The method of claim 49, wherein the RF RX coil network is configured to fully cover the surface such that there is no access aperture on the surface nearest the region of interest.

51. The method of claim 49 or 50, wherein the RF RX coil network is configured for imaging external to the surface by a distance of about 100 mm.

52. The method of any one of claims 49 to 51, wherein the RF RX coil network comprises a plurality of RF RX coils.

53. The method of any one of claims 49 to 52, wherein the RF RX coil network comprises a plurality of interconnected RF RX coils.

54. The method of any one of claims 49 to 53, wherein the RF RX coil network comprises a plurality of coupled RF RX coils.

55. The method of any one of claims 49 to 54, wherein a number of turns and loops of the RF RX coil is configured to be adjustable to cover an entire space between legs of the subject such that the region of interest is entirely or partially covered.

56. The method of any one of claims 49 to 55, wherein the housing further comprises a radio frequency transmit (RF TX) coil proximate to the surface of the housing, wherein the RF TX coil is configured to generate an electromagnetic field in the region of interest.

57. The method of claim 56, wherein the RF TX coil comprises a plurality of figure-8 coils arranged proximal to the surface.

58. The method of claim 56 or 57, wherein the plurality of figure-8 coils are configured to generate a varying magnetic RF field within the region of interest.

59. The method of any one of claims 56 to 58, wherein the plurality of figure-8 coils are orthogonal to each other.

60. The method of any one of claims 56 to 59, wherein the plurality of figure-8 coils are tunable to same radiofrequency (RF) resonant frequencies.

61. The method of any one of claims 56 to 60, wherein the plurality of figure-8 coils are tunable to different RF resonant frequencies.WSGR Docket No. 49880-71960162. The method of any one of claims 56 to 61, wherein the plurality of figure-8 coils are configured to generate a uniform magnetic RF field within the region of interest.

63. The method of any one of claims 56 to 62, wherein the MRI system further comprises an electromagnet configured to generate an electromagnetic field in the region of interest.

64. The method of any one of claims 56 to 63, wherein the housing further comprises a gradient coil set positioned proximate to the surface, wherein the gradient coil set is configured to generate an electromagnetic field in the region of interest.

65. The method of claim 64, wherein the gradient coil set comprises a single-sided gradient coil set.

66. The method of any one of claims 49 to 65, wherein the MRI system is configured to be used for one or more of diagnosis, grading, treatment planning, or monitoring of pelvic conditions.

67. The method of any one of claims 49 to 66, wherein the MRI system comprises a magnetic field strength of less than about 0.5 T.

68. The method of any one of claims 49 to 67, wherein the MRI system comprises one or more of an open or single-sided MRI.

69. The method of any one of claims 49 to 68, wherein the housing comprises a through-bore access aperture.

70. The method of any one of claims 49 to 69, wherein the housing does not comprise a through-bore access aperture.

71. The method of any one of claims 49 to 70, wherein the MRI system is configured to be used in an office setting without shielding or floor reinforcements.

72. The method of any one of claims 49 to 71, wherein the MRI system comprises at least one permanent magnet configured for use without superconducting material.

73. The method of any one of claims 49 to 72, wherein the RF RX coil is configured to capture images of the subject when the subject is in a position in front of or on top of the MRI, wherein the position is a high lithotomy, an inclined lithotomy, or a seated position over the MRI.

74. The method of claim 73, wherein the RF RX coil is configured to capture images of the subject when the subject is in contact with the surface of the MRI in the high lithotomy, the inclined lithotomy, or the seated position.WSGR Docket No. 49880-71960175. The method of claim 73 or 74, wherein the housing is configured to be positioned such that a central axis thereof is perpendicular to a floor when capturing images of the subject in the high lithotomy or the inclined lithotomy.

76. The method of claim 74 or 75, wherein the housing is configured to be positioned such that a central axis thereof is parallel to a floor when capturing images of the subject in the seated position.

77. The method of any one of claims 73 to 76, wherein the MRI system is configured to be usable in a first mode and a second mode, wherein the first mode comprises capturing images of the subject in the high lithotomy or the inclined lithotomy when the housing is positioned such that a central axis thereof is perpendicular to a floor, and wherein the second mode comprises capturing images of the subject in the seated position when the housing is positioned such that a central axis thereof is parallel to the floor.

78. A system for mitigating electromagnetic interference (EMI) in magnetic resonance (MR) imaging, the system comprising:a. at least one computer processor;b. at least one EMI detector configured to obtain EMI data and convey the obtained EMI data to the at least one computer processor;c. a magnetic resonance imaging device configured to image an object within an MR imaging region and obtain MR signal data of the object; andd. one or more non-transitory computer-readable storage media storing instructions that, when executed by the at least one processor, cause the system to:i. build a noise convolution matrix from the obtained EMI data;ii. extract a noise subspace from the noise convolution matrix by computing a low-rank approximation, thereby generating a truncated basis set representative of dominant EMI components;iii. project the noise convolution matrix onto the truncated basis set to generate a reduced-dimensionality noise representation;iv. compute a dynamic transformation individually for each repetition time (TR) interval to map the reduced-dimensionality noise representation to the obtained MR signal data, thereby generating a noise estimate; and v. subtract the noise estimate from the obtained MR signal data to generate corrected MR signal data.

79. The system of claim 78, wherein the at least one EMI detector comprises one or more of: an E-field probe, a surface coil, an electrode connected to the patient's skin, a directWSGR Docket No. 49880-719601connection to a sub-system ground, a direct connection to a patient table, or a 50 Ohm terminated channel.

80. The system of claim 78 or 79, wherein the at least one EMI detector is positioned external to the MR imaging volume.

81. The system of any one of claims 78 to 80, wherein the obtained EMI data comprises data collected using one or more primary MRI receive (Rx) coil elements during a known noise region of an MR pulse sequence.

82. The system of any one of claims 78 to 81, wherein building the noise convolution matrix comprises constructing a multi-dimensional block Hankel convolution matrix using a sliding window operation.

83. The system of any one of claims 78 to 82, wherein the EMI data is obtained from an MR pulse sequence during a predetermined noise-only acquisition region.

84. The system of any one of claims 78 to 83, wherein computing the low-rank approximation comprises applying a dimensionality reduction technique comprising one or more of: singular value decomposition (SVD), principal component analysis (PCA), eigenvalue decomposition, or randomized low-rank matrix approximation.

85. The system of claim 84, wherein generating the truncated basis set comprises thresholding singular values or eigenvalues derived from the dimensionality reduction technique based on a predefined fractional threshold.

86. The system of claim 84 or 85, wherein computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors and Rx coil elements utilizing data acquired from noise-only periods.

87. The system of any one of claims 84 to 86, wherein computing the low-rank approximation comprises applying a noise whitening procedure using all EMI detectors utilizing only data acquired during signal portions of MRI data acquisitions.

88. The system of claim 86 or 87, wherein the noise whitening procedure is applied prior to computing the low-rank approximation using singular value decomposition (SVD).

89. The system of any one of claims 78 to 88, wherein the at least one EMI detector comprises noise coils tuned to frequencies associated with excited bands.WSGR Docket No. 49880-71960190. The system of claim 89, wherein the noise coils are configured to have high sensitivity to selected frequencies, thereby comprising a high Q-factor.

91. The system of claim 89 or 90, wherein the noise coils are tuned over a broader frequency band and comprise a sensitivity versus frequency profile that matches a frequency distribution of an EMI spectrum over the broader frequency band.

92. The system of any one of claims 78 to 91, wherein a noise correlation matrix and subsequent low-rank approximation are re-computed for specified frequency bands having bandwidths equal to a bandwidth of excited slabs, portions of an excited bandwidth corresponding to slices in a z-dimension, or portions dominated by a particular EMI noise source.

93. The system of claim 92, wherein the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only time domain data acquired in a non-excited data acquisition.

94. The system of claim 92, wherein the noise correlation matrix for the specified frequency bands is determined by deploying a Fourier Transform (FT) over noise-only portions of time domain data acquired during an MRI data acquisition.

95. The system of claim 93 or 94, wherein the noise correlation matrix is determined using a sliding window FT on the noise-only time domain data; wherein a sliding window time length is determined by a bandwidth of dominant noise sources or a time length required to obtain statistically significant noise correlation information; and wherein the sliding window FT employs zero filling on negative and positive sides of a time window to improve frequency resolution.

96. The system of any one of claims 78 to 95, wherein a noise correlation computation is applied to data acquired during signal portions of an MR pulse sequence by employing a signal subtraction system to remove MR signal from a signal echo, thereby leaving noise-only data to perform the noise correlation computation using both MRI Rx coil elements and the at least one EMI detector.

97. The system of any one of claims 78 to 96, wherein computing the dynamic transformation comprises fitting the reduced-dimensionality noise representation to the obtainedWSGR Docket No. 49880-719601MR signal data using a least-squares estimation to determine transformation coefficients for each individual TR interval.

98. The system of any one of claims 78 to 97, wherein the instructions are further configured to cause the system to reconstruct an EMI-corrected MR image from the corrected MR signal data.

99. The system of any one of claims 78 to 98, wherein computing the noise estimate and subtracting the noise estimate are performed jointly within an iterative image reconstruction process.

100. The system of claim 99, wherein the iterative image reconstruction process comprises minimizing a joint objective function that simultaneously estimates a reconstructed MR image and the noise estimate.

101. The system of claim 100, wherein the obtained MR signal data comprises k-space data acquired using either a fully sampled acquisition trajectory or an undersampled acquisition trajectory.

102. The system of claim 100 or 101, wherein the joint objective function comprises: a. a data fidelity term that evaluates a difference between the obtained MR signal data and a predicted MR signal derived from the reconstructed MR image and the noise estimate;b. an image regularization term configured to enforce structural properties, on the reconstructed MR image; andc. a noise regularization term configured to enforce a low-rank constraint on the extracted noise subspace.

103. The system of claim 102, wherein the structural properties comprise sparsity or smoothness.

104. The system of claim 102 or 103, wherein the noise regularization term comprises a nuclear norm penalty applied to the noise convolution matrix.

105. The system of any one of claims 102 to 104, wherein minimizing the joint objective function comprises executing an alternating minimization algorithm that alternates between updating the reconstructed MR image and updating the noise estimate.WSGR Docket No. 49880-719601106. The system of claim 105, wherein updating the noise estimate comprises applying singular value thresholding (SVT) to a data residual to dynamically update the truncated basis set at each iteration of the alternating minimization algorithm.

107. The system of any one of claims 102 to 106, wherein the instructions are further configured to cause the system to:a. perform a pre-scan calibration to identify one or more dominant noise frequency bands within the MR imaging environment; andb. adaptively tune the at least one EMI detector to the identified one or more dominant noise frequency bands prior to obtaining the EMI data.

108. The system of claim 107, wherein adaptively tuning the at least one EMI detector comprises applying a hardware or software-based bandpass filter configured to maximize sensitivity to the identified one or more dominant noise frequency bands.

109. The system of any one of claims 102 to 108, wherein the truncated basis set is derived at least in part from a historical noise subspace model, wherein the historical noise subspace model is constructed from accumulated EMI data obtained during a plurality of prior MR imaging sessions at a specific installation site.

110. The system of claim 109, wherein extracting the noise subspace is performed prior to obtaining the MR signal data by computing the low-rank approximation on the accumulated EMI data, thereby generating a pre-computed site-specific basis set that bypasses real-time basis set generation during a current MR imaging session.

111. The system of claim 109 or 110, wherein computing the noise estimate comprises a two-pass estimation process, comprising:a. a first pass that projects the noise convolution matrix onto the historical noise subspace model to remove baseline environmental interference; andb. a second pass that projects a residual noise representation onto a scan-specific basis set derived from the EMI data obtained during the current MR imaging session to remove transient interference.

112. The system of any one of claims 102 to 111, wherein the instructions are further configured to cause the system to:a. compare a rank or a singular value distribution of the extracted noise subspace to a historical baseline noise subspace model;WSGR Docket No. 49880-719601b. determine a deviation metric between the extracted noise subspace and the historical baseline noise subspace model; andc. trigger an automated system action if the deviation metric exceeds a predetermined threshold.

113. The system of claim 112, wherein the automated system action comprises one or more of generating a user alert indicating elevated environmental interference, dynamically increasing a number of signal averages (NEX) for the MR pulse sequence, or switching to an alternative noise cancellation protocol.

114. The system of any one of claims 102 to 113, wherein extracting the noise subspace comprises processing the noise convolution matrix through a trained artificial neural network comprising an undercomplete autoencoder architecture, wherein a bottleneck layer of the undercomplete autoencoder is configured to output a latent representation that defines the truncated basis set.

115. The system of claim 99, wherein the iterative image reconstruction process is executed by a trained unrolled neural network, wherein individual layers or blocks of the trained unrolled neural network correspond to iterations of an alternating minimization algorithm, and wherein the trained unrolled neural network comprises learnable parameters configured to dynamically apply singular value thresholding (SVT) to extract the noise subspace.

116. The system of claim 115, wherein the trained unrolled neural network is trained using a supervised learning protocol comprising:a. a training dataset comprising pairs of EMI-corrupted MR k-space data and corresponding ground-truth unaliased MR images; andb. a loss function configured to minimize a difference between an output of the trained unrolled neural network and the ground-truth unaliased MR images.

117. The system of any one of claims 102 to 116, wherein the MR signal data is acquired using a low-field or ultra-1 ow-fi eld MR scanner operating in an unshielded environment lacking a stationary Faraday cage.

118. The system of any one of claims 102 to 117, wherein the MR signal data is acquired using a single-sided MR scanner.WSGR Docket No. 49880-719601119. The system of any one of claims 102 to 118, wherein the MR signal data is acquired using a portable MR scanner configured for point-of-care operation.

120. The system of any one of claims 102 to 119, wherein the MR signal data is acquired using a pulse sequence comprising non-linear spatiotemporal spatial encoding trajectories.

121. The system of claim 96, wherein the MRI signal used for subtraction is estimated using signals received by a plurality of receive coils and spatial sensitivity relationships between the receive coils.

122. The system of claim 121, wherein estimating the MRI signal includes reconstructing an MRI image and using coil sensitivity profiles to estimate signal contributions in each receive channel.

123. The system of claim 122, wherein the estimated signal contributions are subtracted from the received signals to generate improved estimates of electromagnetic interference present in each receive channel, and wherein the steps of signal estimation, signal subtraction, and noise estimation are repeated iteratively to refine estimates of MRI signal and electromagnetic interference.

124. The system of claim 123, wherein coil sensitivity profiles are estimated or refined using the reconstructed MRI signal and residual signals obtained after signal subtraction.

125. The system of claim 123 or 124, wherein spatial signatures of electromagnetic interference across the receive coils are estimated and used to suppress interference components.

Images