Patent Application
20110093233
A portion of the disclosure of this patent document contains material that is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction of the patent document or the patent disclosure as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
Conventional generalized auto-calibrating partially parallel acquisitions (GRAPPA) generates uncombined coil images for each coil in an array of receive coils used by a parallel magnetic resonance imaging (pMRI) apparatus. GRAPPA reconstructs missing lines in each coil element by forming linear combinations of neighboring lines to reconstruct individual missing data points. The weights for these linear combinations are derived by forming a fit between additionally acquired lines using a pseudo-inverse operation. GRAPPA is described in Griswold, et al., Proceedings of the ISMRM, Vol. 7, Issue 6, Pg. 1202-1210 (2002).
Conventional Radial GRAPPA acquires data and makes a reconstruction kernel comprised of GRAPPA weights. The reconstruction kernel is used to reconstruct rays acquired during a radial reconstruction. The quality of a radial GRAPPA reconstruction depends, at least in part, on whether a suitable reconstruction kernel that corresponds to a ray being reconstructed is available. Radial GRAPPA is described in Griswold et al., Proc. ISMRM 11, 2003, p2349.
An under-sampled radial acquisition will not acquire every possible ray in a radial pattern. Assuming that 360 rays are available, one for each degree in a circle associated with a radial pattern, a fully sampled data set would acquire a ray at multiple rotations (e.g., 0 degrees, 1 degree, 2 degrees). However, in an under-sampled radial acquisition, less than every ray will be acquired. For example, rays may be acquired at 0 degrees, 2 degrees, 4 degrees, and so on. Therefore there are rays missing at 1 degrees, 3 degrees, and so on. However, these missing rays can be filled in using conventional techniques to produce acceptable results.
An acknowledged but tolerable error associated with conventional reconstruction assumes that a ray for 0 degrees is useful for reconstructing a neighboring ray at, for example 1 degree. This assumption holds for rays that are closely spaced as is the case when relatively mild under-sampling factors are used. However, when the two rays get too far apart (e.g., 0 degrees and ten degrees) this assumption and reconstruction approach no longer produces acceptable results. Also, the assumption is weaker at the edge of k-space where points on rays are farther apart than at the center of k-space where points on rays intersect or nearly overlap. The review of GRAPPA provided below facilitates understanding this neighbor based approach and its shortcomings at high acceleration factors.
To better understand radial GRAPPA and the example systems and methods described below, a brief history of GRAPPA, starting at SMASH (Simultaneous Acquisition of Spatial Harmonics) is provided.
Performing the reconstruction requires determining the weights to be used in the reconstruction. As in VD-AUTO-SMASH, a block of extra ACS lines is acquired in the center of k-space and used to determine the complex weights.
Conventional parallel imaging techniques may fill in omitted k-space lines prior to Fourier transformation by constructing a weighted combination of neighboring lines acquired by the different RF detector coils. Conventional parallel imaging techniques may also first Fourier transform under-sampled k-space data set to produce an aliased image from each coil and then unfold the aliased signals by a linear transformation of the superimposed pixel values.
Non-Cartesian imaging has advantages over standard Cartesian imaging due to, for example, efficient k-space coverage or suppression of off-resonance effects. However, points acquired in a non-Cartesian approach do not necessarily fall onto a grid and thus have conventionally been re-sampled onto a Cartesian matrix before a Fourier transform is performed. One example gridding technique is the self-calibrating GRAPPA operator gridding (GROG) method. Using GROG, non-Cartesian MRI data is gridded using spatial information from a multichannel coil array without an additional calibration dataset. Using self-calibrating GROG, the non-Cartesian data points are shifted to nearby k-space locations using parallel imaging weight sets determined from the data points themselves. GROG employs the GRAPPA Operator, a special formulation of the general reconstruction method GRAPPA, to perform these shifts. While this re-gridding produces acceptable results in radial trajectories at low acceleration factors, at higher acceleration factors it may yield sub-optimal results.
Re-gridding has been employed in Radial GRAPPA, (Griswold, et al., “Direct Parallel Imaging Reconstruction Of Radially Sampled Data Using GRAPPA With Relative Shifts.,” Proceedings of the ISMRM 11th Scientific Meeting, Toronto, 2003: 2349). Radial GRAPPA improves on conventional pMRI processing using non-Cartesian trajectories. Recall that GRAPPA determined a linear combination of individual coil data to create missing lines of k-space. GRAPPA determined the coefficients for the combination by fitting the acquired data to some over-sampled data near the center of k-space. The over-sampled data is acquired using ACS lines.
With conventional radial GRAPPA, a preliminary fully sampled scan is first performed to acquire training data that is used to estimate the missing radial data. This training data can then be used throughout a real-time scan to estimate radial lines that were not sampled. In order to calculate the required weights, multiple points in the region are used together to solve for the required number of unknown weights. Given a typical number of unknowns (e.g., 240 unknowns), a typical region size could include 8 rays and 32 points along the ray. The configuration of the different points is assumed to be the same within each region. A weight set is then derived for each region and the reconstruction is performed region by region. Note that this weight solution is the best fit solution for all of the points in the region, which is in effect correct only for the average point configuration in the region. In practical implementations, this means that some level of error is distributed to every reconstruction in the region. In addition, because only a single fully sampled data set is used for calibration, and because of its intrinsic sensitivity to variations in the point structure within each region, conventional radial GRAPPA has relied on high quality fully sampled training data that may have required extensive signal averaging. Conventionally, this acquisition may have been impractical for certain applications (e.g., contrast enhanced dynamic studies). Additionally, the errors resulting from too widely separated acquired rays has limited the maximal undersampling possible with radial GRAPPA.
The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate various example systems, methods, and other example embodiments of various aspects of the invention. It will be appreciated that the illustrated element boundaries (e.g., boxes, groups of boxes, or other shapes) in the figures represent one example of the boundaries. One of ordinary skill in the art will appreciate that in some examples one element may be designed as multiple elements or that multiple elements may be designed as one element. In some examples, an element shown as an internal component of another element may be implemented as an external component and vice versa. Furthermore, elements may not be drawn to scale.
Example systems and methods acquire calibration data at different points in time and perform a through-time calibration for radial GRAPPA. The calibration data may be fully sampled calibration sets but may also be less than fully sampled calibration data sets. By acquiring calibration data through time, multiple copies of each point can be acquired. Using these multiple copies, one can derive a separate reconstruction kernel for each desired reconstruction point in the raw data. Because an exact kernel configuration can be calculated for each point, the resulting reconstruction kernel will support higher acceleration factors for under-sampling than previously thought possible for radial GRAPPA. The radial calibration data is acquired according to a plan that acquires radial rays that are in the same configuration as rays that will be used in a reconstruction. Since data is acquired through time, the reconstruction kernel may be exact for the rays that are acquired multiple times through time.
By repeatedly acquiring calibration data for a ray throughout a period of time, a point in k-space to be solved for using the reconstruction kernel can be successfully reconstructed based on the high quality calibration data. Consider a calibration data set that acquires a calibration line for 0 degrees and for 5 degrees at several points in time. At each point in time there will be a ray for zero degrees and a ray for five degrees. While the calibration data set need not be fully sampled, it will be configured to have the same configuration as the reconstruction kernel. This means that if a reconstruction will rely on rays for 0 degrees, 5 degrees, 10 degrees, . . . , then the calibration data set will acquire, through time, multiple copies of calibration data for the reconstruction kernel rays. The reconstruction kernel constructed from these repeatedly acquired rays can be very accurate.
At high acceleration factors (e.g., R>4), conventional radial GRAPPA may experience artifacts that render an image substantially unusable. Recall that in radial GRAPPA, segments of radial k-space can be approximated as Cartesian segments during parallel imaging reconstruction. However, at large acceleration factors the Cartesian approximation may not hold up, yielding the unacceptable artifacts.
The following includes definitions of selected terms employed herein. The definitions include various examples and/or forms of components that fall within the scope of a term and that may be used for implementation. The examples are not intended to be limiting. Both singular and plural forms of terms may be within the definitions.
References to “one embodiment”, “an embodiment”, “one example”, “an example”, and so on, indicate that the embodiment(s) or example(s) so described may include a particular feature, structure, characteristic, property, element, or limitation, but that not every embodiment or example necessarily includes that particular feature, structure, characteristic, property, element or limitation. Furthermore, repeated use of the phrase “in one embodiment” does not necessarily refer to the same embodiment, though it may.
“Computer-readable medium”, as used herein, refers to a medium that stores signals, instructions and/or data. A computer-readable medium may take forms, including, but not limited to, non-volatile media, and volatile media. Non-volatile media may include, for example, optical disks, magnetic disks, and so on. Volatile media may include, for example, semiconductor memories, dynamic memory, and so on. Common forms of a computer-readable medium may include, but are not limited to, a floppy disk, a flexible disk, a hard disk, a magnetic tape, other magnetic medium, an ASIC, a CD, other optical medium, a RAM, a ROM, a memory chip or card, a memory stick, and other media from which a computer, a processor or other electronic device can read.
“Logic”, as used herein, includes but is not limited to hardware, firmware, software in execution on a machine, and/or combinations of each to perform a function(s) or an action(s), and/or to cause a function or action from another logic, method, and/or system. Logic may include a software controlled microprocessor, a discrete logic (e.g., ASIC), an analog circuit, a digital circuit, a programmed logic device, a memory device containing instructions, and so on. Logic may include one or more gates, combinations of gates, or other circuit components. Where multiple logical logics are described, it may be possible to incorporate the multiple logical logics into one physical logic. Similarly, where a single logical logic is described, it may be possible to distribute that single logical logic between multiple physical logics.
“Signal”, as used herein, includes but is not limited to, electrical signals, optical signals, analog signals, digital signals, data, computer instructions, processor instructions, messages, a bit, a bit stream, or other means that can be received, transmitted and/or detected.
Example systems and methods control a parallel magnetic resonance imaging (pMRI) apparatus to acquire a set of radial calibration data and to perform a through-time calibration based, at least in part, on the set of radial calibration data. In one example, the radial calibration data may be fully sampled. However, a fully sampled data set is not required. Example systems and methods control the pMRI apparatus to acquire multiple data sets, where a data set will have at least the same rays that will be used in a reconstruction kernel. For example, if a reconstruction kernel is going to rely on data for rays at 0 degrees, 5 degrees, 10 degrees, 15 degrees, and so on, then multiple radial data sets that include data on at least those rays will be acquired. The calibration data sets will be acquired at different points in time.
In one embodiment, since the multiple radial data sets are acquired at multiple points in time, the calibration data used to build the reconstruction kernel can be very accurate. At each point in time there will be a radial calibration data set for each ray used in the reconstruction kernel. This facilitates creating an improved reconstruction kernel having improved GRAPPA weights, which in turn facilitates reducing artifacts in reconstructions of highly under-sampled radial GRAPPA. Returning to the example above, if 240 weights are required, then instead of assembling at least 240 different points in the region to solve for the weights, assemble 240 separate acquisitions in time. In each case, the different points would then be assembled into a set of linear equations that describe the relationship between different acquired points and a potential reconstructed points. The system of equations is then solved for the required weights. The difference between the two methods is that the weights would be in error for each of the reconstructed points in the former region-based case. By contrast, example apparatuses and methods described herein facilitate deriving an exact set of weights using through-time calibration.
Performing a through-time time calibration could also be referred to as calibrating the pMRI with a set of calibration data acquired at different points over a period of time.
In one embodiment, the through-time calibration facilitates producing weights 330 for specific locations in k-space. The weights may be computed from radial calibration data sets 310, 312, 314, 316, and 318. While five radial calibration data sets are illustrated, one skilled in the art will appreciate that a greater and/or lesser number of radial calibration data sets may be employed. The weights may then be employed to reconstruct under-sampled data sets 322, 324, and 326. In
Data associated with a reconstructed image, with GRAPPA weights employed for computing a reconstructed image, and with calibration data associated with computing the GRAPPA weights can be stored on a computer-readable medium. The reconstructed image represents items including, for example, human bones, human tissues, human blood, and so on. In one example, a computer-readable medium may store, in a first field, data representing a radial calibration data set acquired by a pMRI apparatus. The radial calibration data set is acquired from an object to be imaged (e.g., heart, knee, lung, vasculature). The computer-readable medium may also store, in a second field, data representing an under-sampled radial data set acquired by the pMRI apparatus. The under-sampled radial data set is also acquired from a real-world physical object (e.g., heart, lung). The computer-readable medium may also store, in a third field, data representing GRAPPA weights calibrated for a point missing in the under-sampled radial data set. The GRAPPA weights in the third field are computed from data in the first field and are applied to data in the second field.
MRI apparatus 500 may include a set of RF antennas 550 that are configured to generate RF pulses and to receive resulting magnetic resonance signals from an object to which the RF pulses are directed. In some examples, how the pulses are generated and how the resulting MR signals are received may be controlled and thus may be selectively adapted during an MRI procedure. Separate RF transmission and reception coils can be employed. The RF antennas 550 may be controlled, at least in part, by a set of RF transmission units 560. An RF transmission unit 560 may provide a signal to an RF antenna 550.
The gradient coils supply 540 and the RF transmission units 560 may be controlled, at least in part, by a control computer 570. In one example, the control computer 570 may be programmed to control a pMRI device as described herein. The magnetic resonance signals received from the RF antennas 550 can be employed to generate an image and thus may be subject to a transformation process. The transformation can be performed by an image computer 580 or other similar processing device. The image data may then be shown on a display 590. While
Apparatus 599 also includes an under-sampling acquisition logic 620. Under-sampling acquisition logic 620 is configured to acquire an under-sampled radial data set from the object to be imaged. Due to the through-time radial GRAPPA calibration described herein, under-sampling acquisition logic 620 can acquire under-sampled radial data sets using a larger acceleration factor (e.g., R=8) than is conventionally possible.
Apparatus 599 also includes a through-time radial GRAPPA calibration logic 630. Through-time radial GRAPPA calibration logic 630 is configured to compute a GRAPPA weight set for a point missing from k-space in the under-sampled radial data set. The GRAPPA weight set is calibrated for the missing point and computed from data in the plurality of radial calibration data sets. In one embodiment, the through-time radial GRAPPA calibration logic 630 is configured to compute a value for each point missing from k-space in the under-sampled radial data set using a GRAPPA weight set calibrated and applied for each missing point. In one embodiment, the GRAPPA weight set is computed from data selected from each member of the plurality of radial calibration data sets. In another embodiment, the GRAPPA weight set is computed from data selected from less than each member of the plurality of radial calibration data sets. One skilled in the art will appreciate that in different embodiments different calibration data can be used to compute the GRAPPA weight set.
The radial dataset acquisition logic 610 is configured to acquire radial calibration data sets comprising two or more rays for which calibration data is acquired. In one example, one of the rays is a ray that would be acquired in the under-sampled data, while the other could be one that would be skipped in the under-sampled acquisition. Thus the two or more rays are selected based on rays that will be used to reconstruct the image. The two or more rays will be used by the reconstruction logic 640 to reconstruct the image and that will be used by the through-time radial GRAPPA calibration logic 630 to compute the GRAPPA weight set. In different embodiments a radial calibration data set may include 12, 24, 48 and other numbers of rays. In some embodiments the rays may be evenly spaced, while in other embodiments the rays may not be evenly spaced.
In one embodiment, the through-time radial GRAPPA calibration logic 630 can be configured to compute the GRAPPA weight set from all the rays acquired in the radial calibration data sets. In another embodiment, the through-time radial GRAPPA calibration logic 630 can be configured to compute the GRAPPA weight set from less than all the rays acquired in the radial calibration data sets. In different embodiments the radial calibration data sets can be fully sampled data sets or less than fully sampled data sets. In different embodiments, through-time calibration can also be based, at least in part, on a small amount of conventional region based calibration.
Apparatus 599 also includes a reconstruction logic 640. Reconstruction logic 640 is configured to reconstruct an image from the under-sampled radial data set. The reconstruction will depend, at least in part on the GRAPPA weight sets.
Some portions of the detailed descriptions that follow are presented in terms of algorithms and symbolic representations of operations on data bits within a memory. These algorithmic descriptions and representations are used by those skilled in the art to convey the substance of their work to others. An algorithm, here and generally, is conceived to be a sequence of operations that produce a result. The operations may include physical manipulations of physical quantities. Usually, though not necessarily, the physical quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated in a logic, and so on. The physical manipulations create a concrete, tangible, useful, real-world result.
It has proven convenient at times, principally for reasons of common usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, and so on. It should be borne in mind, however, that these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise, it is appreciated that throughout the description, terms including processing, computing, determining, and so on, refer to actions and processes of a computer system, logic, processor, or similar electronic device that manipulates and transforms data represented as physical (electronic) quantities.
Example methods may be better appreciated with reference to flow diagrams. While for purposes of simplicity of explanation, the illustrated methodologies are shown and described as a series of blocks, it is to be appreciated that the methodologies are not limited by the order of the blocks, as some blocks can occur in different orders and/or concurrently with other blocks from that shown and described. Moreover, less than all the illustrated blocks may be required to implement an example methodology. Blocks may be combined or separated into multiple components. Furthermore, additional and/or alternative methodologies can employ additional, not illustrated blocks.
Method 700 also includes, at 720, controlling the pMRI apparatus to acquire an under-sampled radial data set from the object to be imaged. One skilled in the art will appreciate that there are different methods for acquiring an under-sampled radial data set.
Method 700 also includes, at 730, controlling the pMRI apparatus to perform a through-time radial GRAPPA calibration. The through-time radial GRAPPA calibration includes computing a GRAPPA weight set from data in the two or more calibration data sets. In one example, the through-time radial GRAPPA calibration includes computing a value for each point missing from k-space in the under-sampled radial data set using a GRAPPA weight set calibrated and applied for each missing point. In different examples, the GRAPPA weight set can be computed from data selected from each of the two or more radial calibration data sets and/or from less than each of the two or more radial calibration data sets. For example, radial calibration data sets may be selected based on proximity in time to an under-sampled data set, based on a sliding window of time in which radial calibration data sets are acquired, based on complete coverage, and so on.
Method 700 also includes, at 740, controlling the pMRI apparatus to reconstruct an image of the object to be imaged from the under-sampled radial data set. A value for a point missing from k-space in the under-sampled radial data set is computed using the GRAPPA weight set as calibrated and applied for the missing point. In one example, the image can be reconstructed in real-time. Real-time reconstruction is useful in applications where the object to be imaged is, for example, a beating heart, a lung, a region of a human vasculature in which blood is flowing, and so on. In these examples, the two or more radial calibration data sets can be acquired from the object to be imaged at different points in time throughout a period of time during which the object to be imaged moves. Unlike conventional systems that rely on additional external timing, in method 700 the radial calibration data sets are acquired from the object to be imaged without reference to an EKG gating signal and/or while the object to be imaged is breathing normally without breath-holding. In different embodiments, the calibration data and the under-sampled data can be acquired in different ways. For example, method 700 can include controlling the pMRI apparatus to acquire all the radial calibration data sets and then to acquire the under-sampled radial data set or to interleave acquisition of the radial calibration data sets and the under-sampled radial data set.
While
In one example, a method may be implemented as computer executable instructions. Thus, in one example, a computer-readable medium may store computer executable instructions that if executed by a machine (e.g., processor) cause the machine to perform method 700. While executable instructions associated with the method 700 are described as being stored on a computer-readable medium, it is to be appreciated that executable instructions associated with other example methods described herein may also be stored on a computer-readable medium.
While example systems, methods, and so on have been illustrated by describing examples, and while the examples have been described in considerable detail, it is not the intention of the applicants to restrict or in any way limit the scope of the appended claims to such detail. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the systems, methods, and so on described herein. Therefore, the invention is not limited to the specific details, the representative apparatus, and illustrative examples shown and described. Thus, this application is intended to embrace alterations, modifications, and variations that fall within the scope of the appended claims.
To the extent that the term “includes” or “including” is employed in the detailed description or the claims, it is intended to be inclusive in a manner similar to the term “comprising” as that term is interpreted when employed as a transitional word in a claim.
To the extent that the term “or” is employed in the detailed description or claims (e.g., A or B) it is intended to mean “A or B or both”. When the applicants intend to indicate “only A or B but not both” then the term “only A or B but not both” will be employed. Thus, use of the term “or” herein is the inclusive, and not the exclusive use. See, Bryan A. Garner, A Dictionary of Modern Legal Usage 624 (2d. Ed. 1995).
To the extent that the phrase “one or more of, A, B, and C” is employed herein, (e.g., a data store configured to store one or more of, A, B, and C) it is intended to convey the set of possibilities A, B, C, AB, AC, BC, and/or ABC (e.g., the data store may store only A, only B, only C, A&B, A&C, B&C, and/or A&B&C). It is not intended to require one of A, one of B, and one of C. When the applicants intend to indicate “at least one of A, at least one of B, and at least one of C”, then the phrasing “at least one of A, at least one of B, and at least one of C” will be employed.
