Patent Application
20150305645
The subject matter below relates generally to magnetic resonance imaging (MRI). In particular, the subject matter relates to arterial spin labeling (ASL) and perfusion MRI.
The MRI system shown in
An MRI system controller 22 has input/output ports connected to a display 24, keyboard 26 and printer 28. As will be appreciated, the display 24 may be of the touch-screen variety so that it provides control inputs as well and a mouse or other I/O device(s) may be provided.
The MRI system controller 22 interfaces with MRI sequence controller 30 which, in turn, controls the Gx, Gy and Gz gradient coil drivers 32, as well as the RF transmitter 34 and the transmit/receive switch 36 (if the same RF coil is used for both transmission and reception). The MRI sequence controller 30 includes suitable program code structure 38 for implementing MRI imaging (also known as nuclear magnetic resonance, or NMR, imaging) techniques, which may also include parallel imaging. As described below, sequence controller 30 may be configured to apply a predetermined tagging pulse sequence and a predetermined control pulse sequence, in order to obtain corresponding tagging and control images from which a diagnostic MRI image is obtained. MRI sequence controller 30 may also be configured for EPI imaging and/or parallel imaging. Moreover, MRI sequence controller 30 may facilitate one or more preparation scan (prescan) sequences, and a scan sequence to obtain a main scan MR image (sometimes referred to as a diagnostic image).
The MRI system 20 includes an RF receiver 40 providing input to data processor 42 so as to create processed image data, which is sent to display 24. The MRI data processor 42 is also configured for access to previously generated MR data, images, and/or maps, and/or system configuration parameters 46 and MRI image reconstruction program code structures 44 and 50.
Also illustrated in
Indeed, as those in the art will appreciate, the
Not only does the physical state of processing circuits (e.g., CPUs, registers, buffers, arithmetic units, etc.) progressively change from one clock cycle to another during the course of operation, the physical state of associated data storage media (e.g., bit storage sites in magnetic storage media) is transformed from one state to another during operation of such a system. For example, at the conclusion of an image reconstruction process and/or sometimes the generation of a subtracted image from control and tagging images, as described below, an array of computer-readable accessible data value storage sites in physical storage media will be transformed from some prior state (e.g., all uniform “zero” values or all “one” values) to a new state wherein the physical states at the physical sites of such an array vary between minimum and maximum values to represent real world physical events and conditions (e.g., the internal physical structures of a patient over an imaging volume space). As those in the art will appreciate, such arrays of stored data values represent and also constitute a physical structure—as does a particular structure of computer control program codes that, when sequentially loaded into instruction registers and executed by one or more CPUs of the MRI system 20, causes a particular sequence of operational states to occur and be transitioned through within the MRI system.
Arterial spin labeling (ASL) is an MRI technique that is of particular interest for perfusion and non-contrast enhanced MRA applications. ASL relies upon the inflow of blood into the volume being imaged, and uses separate control and tag pulse sequences to label (i.e., tag) spins of inflowing blood differently. Separate images are generated based upon the control pulse sequence and the tag pulse sequence. An image generated based upon a control pulse sequence is referred to as a “control image,” and an image generated based upon a tag pulse sequence is referred to as a “tag image.” A perfusion MRA image can be obtained by subtracting the tag image from the control image.
Dai et al., “Continuous Flow-Driven Inversion for Arterial Spin Labeling Using Pulsed Radio Frequency and Gradient Fields,” Magnetic Resonance in Medicine 60:1488-1497 (2008), describes pseudo-continuous arterial spin labeling (pCASL) which is used frequently for many applications including intracranial applications. However, its tagging efficiency is highly sensitive to off-resonance effects and gradient imperfections, which induce phase mismatches or phase errors between the radiofrequency pulses (Wu et al., Magnetic Resonance in Medicine 58:1020-27 (2007)). This sensitivity can lead to tagging efficiency loss, signal to noise ratio (SNR) loss, and unpredictable variations in acquired perfusion images. The high sensitivity may be due, at least in part, to the tag and control conditions of flowing arterial blood being, to a significant extent, defined by the specification of the phases in the RF pulse train. Jung et al., “Multiphase Pseudocontinuous Arterial Spin Labeling (MP-PCASL) for Robust Quantification of Cerebral Blood Flow,” Magnetic Resonance in Medicine 64:799-810 (2010), described a variation of pCASL that may have reduced the sensitivity to off-resonance artifact.
Regional Perfusion Imaging (RPI) based on ASL provides the ability to noninvasively delineate the perfusion territories of major cerebral arteries. Rather than injecting a flow tracer, ASL employs RF and magnetic field gradient pulses to invert naturally existing water spins in the feeding arteries. Many ASL techniques including pCASL and MP-PCASL noted above, however, label all the arteries feeding the perfusion region.
Several ASL techniques have been proposed for observing individual perfusion territories. The general principle of these RPI techniques is to tag only arterial spins flowing through the artery or arteries of interest, while avoiding the tagging of spins in other arteries. Control over which arteries are labeled can be used to measure the tissue regions that are perfused by particular vessels (e.g., arteries) and to characterize the dynamics of flow through vessels, occlusions, arteriovenous malformations, aneurysms, and the like.
In some applications, the delineation of perfusion territories by RPI provides complementary information to angiography, such as, for example, information regarding the status of blood flow in different regions of the arterial tree.
Dai et al., “Modified Pulsed Continuous Arterial Spin Labeling for Labeling a Single Artery,” Magnetic Resonance in Medicine 64:975-982, 2010 (hereafter “Dai VS-pCASL”) which is herein incorporated by reference in its entirety, describes one or more techniques for modifying pCASL RF pulse sequences to selectively map vascular territories of major cerebral feeding arteries. In the vessel-selective, or single-artery, pCASL approach (VS-pCASL), RPI is accomplished by inserting additional in-plane gradients in the gaps between discrete RF pulses to modulate the phases of flowing spins in different vessels in the labeling plane.
RPI can be a very useful clinical tool to investigate several cerebrovascular disorders or diseases, such as, for example, occlusion in internal carotid arteries (ICAs) (Hendrikse et al., Neurosurgery 57:486-96 (2005); van Laar et al., Radiology 242:526-34 (2007)), arteriovenous malformations (Fiehler et al., AJNR Am J Neuroradiol 30:356-61 (2009)), and collateral flow between major arteries (Hendrikse et al., Stroke 35:882-7 (2004)).
However, similar to pCASL, VS-pCASL too is vulnerable to off-resonance effects, which can cause degradation in vessel-selective tagging efficiency and failure in vessel-selective perfusion imaging. Consequently, this sensitivity may compromise the application of VS-pCASL in clinical settings.
Loss in vessel-selective tagging efficiency can be especially true in experiments to separate perfusion regions of left, right ICAs and vertebral artery, where off-resonance effects (e.g., strong field inhomogeneities) are usually observed at the labeling plane around the neck. Such inhomogeneities can also be a concern when imaging perfusion territories of smaller arterial branches, such as Circle-of-Willis (COW) branches. For example, in some brain regions, such as the orbital frontal cortex, significant magnetic field inhomogeneity artifacts exist due to their close proximity of tissue/air boundaries (Truong et al., “Three-dimensional numerical simulations of susceptibility-induced magnetic field inhomogeneities in the human head,” Magnetic Resonance in Medicine 20:759-70 (2002)). The vessel-selective labeling can be seriously contaminated if the labeling plane of the target artery (for example, arterial cerebral arteries) passes through these regions.
A careful manual shimming before VS-pCASL tagging may improve the main field homogeneity and lessen the influence of off-resonance effects; however, in practice, sufficient field homogeneity cannot be achieved by shimming alone.
In short, the single-artery, or vessel-selective, pCASL sequence has been demonstrated to provide regional perfusion maps non-invasively. However, similar to the original pCASL labeling, vessel-selective pCASL is also observed to be vulnerable to off-resonance effects, which introduce phase errors in the labeling RF train and thus cause degradation in tagging efficiency. Below, we propose to restore the signal loss due to off-resonance artifacts by applying a modified multiple phase correction method in the vessel-selective labeling sequence.
Embodiments described in the present application include novel schemes to restore the signal loss due to off-resonance artifacts by applying a phase correction technique in the VS-pCASL labeling sequence or other territory-selective ASL-based sequences. Embodiments provide for estimating the phase offsets or phase errors at the target feeding artery, and effectively restoring the corresponding signal loss due to off-resonance artifact. In this manner, some embodiments provide higher SNR and more robust measurements in VS-pCASL or other territory-selective ASL-based sequences.
As illustrated, the tagging pulse train comprises equally-spaced RF pulses. A small imbalance in the gradients along the flow direction is added for tagging. In the control pulse train, the RF pulses are equally-spaced but maintain a 180-degree phase shift between consecutive pulses.
Specifically, in order to achieve vessel-selectivity, in addition to the labeling gradient along the flow direction as used in pCASL, VS-pCASL introduces in-plane gradients between the RF pulses which produce a phase shift between vessels. The direction of the in-plane gradients is then rotated as illustrated in
VS-pCASL results in the selective labeling of a disk, the center of which is on a target vessel. The center of the disk is controlled by the phases of the RF pulses. The phase of each RF pulse is incremented in phase relative to the pulse immediately before it by an angle determined based upon the applied gradients and the desired disk center. Dai VS-pCASL provides techniques for calculating the phases for VS-pCASL pulse sequence.
At operation 302, process 300 for off-resonance correction of ASL-based perfusion images is entered. The MRI system and the patient are then, at operation 304, prepared for scanning. Operation 304 may include positioning the patient and/or the part of the patient to be imaged in relation to transmit and/or receive coils of the MRI system, and setting of general parameters and/or configuration options for performing imaging.
The techniques described herein can be applied to image many parts of the patient, such as, but not limited to, head, neck, knee, or other area, with appropriate configurations of the system and positioning of the patient. As described below, certain configurations, such as, for example, tagging and/or control slab locations, tagging slice thickness, the number of tagging pulses, a total duration of tagging, and time delay between tagging pulses can be adjusted in a respective manner based upon selected characteristics of the object image. For example, configurations may be set and/or adjusted in accordance with the flow speed of the vessel or specific part of the body or organ being imaged. Other configurations may include specifying a vessel or vessels (e.g., in a head or neck scanning application, the left or right ICA) in which the blood is to be tagged.
The preparation stage may, in some embodiments, also include acquiring one or more prescans, for example, to obtain one or more low resolution MRI images for positioning the patient, coil calibration, locating tagging and/or control slabs/planes, and/or to determine the position of the vessel(s) identified for tagging.
At operation 306, the inversion response of the ASL technique for the target vessel as a function of phase offset is simulated. The “inversion response” represents the ratio of the net magnetization along the z-axis (e.g., obtained by subtracting control image—tag image) to the magnetization of relaxed blood. Simulations may be performed to obtain values for the inversion responses of VS-pCASL labeling at the target vessel as a function of phase offset (i.e., ALP discussed below). Simulated inversion responses from an example simulation are shown as small circles in
The simulation may be provided with initial parameters for properties (e.g., shape, width, spacing between pulses, amplitude of pulses, number of pulses, flip angle, phase, etc.) for tagging and control RF pulses, gradient parameters (e.g., Gx, Gy, Gz, average gradient strength for each gradient, amplitude, etc.). Other parameters may also include tagging plane, control plane, vessel(s) to be tagged, and imaging plane configurations. Yet other parameters provided may include in-plane gradient rotation rates, and in-plane gradient rotation pattern (e.g., Gx as a particular sine curve and Gy as a particular cosine curve) which may be used for vessel-selective tagging.
According to an embodiment, numerical Bloch simulations are performed to determine the simulated inversion responses (control-tag) of the vessel-selective pCASL labeling at the target vessel as a function of phase offset (Δψ as shown in
At operation 308, the simulated inversion responses are fitted to a polynomial. According to an embodiment, the simulated inversion response curve was fitted to a 12th order polynomial P(Δψ), shown in
At operation 310, the polynomial is stored to be subsequently used as the signal model for the correction process where the perfusion signal can be estimated by fitting the measured perfusion-weighted data (mi,n) at multiple phase offsets to the expected inversion efficiency function in a voxel-by-voxel manner as shown in equation (1) where CBF is the perfusion-weighted map, and ε is the phase error map. In embodiments, the availability of a polynomial, or more specifically a high order polynomial such as, but not limited to, a 12th order polynomial, as a signal model may provide a better fit than other types of functions that may be fitted to the simulated data points, as more free variables are available in the polynomial fit.
Operation 306 may or may not be performed during the scanning process. In some embodiments, the simulation may be performed entirely, or in part, offline from the scanning, and the results uploaded to the MRI system. In other embodiments, the simulation may be performed on-line, for example, by accessing configuration parameters (e.g., pulse configurations, vessel selection etc.) automatically, either during or after the preparation processing of operation 304.
At operation 312, the RF pulse sequence with off-resonance correction is applied. According to an embodiment, the applied RF pulse sequence is the VS-pCASL pulse sequence modified to include off-resonance correction. The original VS-pCASL is illustrated in
Returning to
Operation 312 may also include acquisition of images. In some embodiments, the tagging pulse sequence and the control pulse sequence are each configured with an imaging pulse train. Thus, in some embodiments, the tagging pulse sequence includes a train of tagging pulses and a train of imaging pulses; and the control pulse sequence includes a train of control pulses and a train of imaging pulses.
According to an embodiment, an imaging sequence follows each tagging pulse train and each control pulse train. The imaging may be performed according to a predetermined imaging pulse such as, but not limited to, 2D/3D Field Echo (FE), Fast Field Echo (FFE), Fast Spin Echo (FSE), Steady State FSE (SSFSE), Balanced Steady-State Free Precession (bSSFP), Ultrashort Echo Time (UTE), etc., imaging pulse sequences. In one or more embodiments, the imaging pulse trains in the tagging pulse sequence and the control pulse sequence may be identical.
Images may be acquired for a plurality of phase offsets. In an embodiment, separate images are acquired for phase offsets Δψ=−120°, −90°, −60°, −30°, 0°, 30°, 60°, 90°, and 120°. Acquiring an image at a particular phase offset may include acquisition of corresponding tag and control images, and the subtraction of the tag from the control image. In one or more embodiments, a predetermined number (e.g., the number of unique phase offsets configured for imaging) of separate images are acquired with each image corresponding to a unique phase offset. The acquired images represent uncorrected perfusion-weighted images.
At operation 314, the measured perfusion data is fitted to the signal model discussed in relation to operation 310. The measured perfusion data is obtained from the uncorrected perfusion-weighted images. The fitting of the measured perfusion data to the signal model may be performed on a pixel-by-pixel basis of a yet to be created corrected perfusion-weighted image. Operation 316 may be considered a post-processing activity to the extent that it is performed after the NMR data acquisition has been completed.
The curve fitting at operation 314 may include, for each voxel i in the yet to be formed corrected perfusion-weighted image, having a plurality of measured values mi,n where n ranges from 1 to the number of images acquired. According to an embodiment, a separate image is acquired for each unique phase offset from a predetermined set of phase offsets. For example, separate images may be acquired for each of −120°, −90°, −60°, −30°, 0°, 30°, 60°, 90° and 120° phase offsets, yielding a total of nine images that can contribute to voxel i, which may be represented as mi, n where n=1 . . . 9. In this example, the curve fitting includes fitting mi, n to the signal model. The curve fitting may be performed by any appropriate curve fitting technique. According to an embodiment, the curve fitting may be performed according to a minimum root-mean-square error technique.
Due to off-resonance artifact in the labeling plane and/or arterial blood, measured values mi,n, may be shifted in phase relative to the signal model. The amount of phase shift or phase offset of mi,n from the signal model for i may be represented as εi.
The measured perfusion-weighted data mi,n may be represented as in the following equation (1): mi,n=CBFi×P(Δψn−εi), where P is the signal model, and εi is the phase error or phase offset determined by the curve fitting. CBFi is corrected cerebral blood flow value for voxel i (also referred to as a corrected perfusion-weighted map value). Then, because mi,n is known from the measurement and P(Δψn−εi) is known from the fitted curve, CBFi and εi can be determined from equation (1). It should be noted that, after the curve fitting of mi,n is performed, CBFi is not dependent on either the number of phase offsets or the measured value at a particular phase offset.
At operation 316, the corrected perfusion-weighted image (CBFi) is generated, and at operation 318, the obtained corrected perfusion-weighted image may be output to a display, to storage, directed to a printer, or communicated to another device for further processing. According to an example embodiment, the corrected perfusion-weighted image may be used to view a tissue region of interest in which the perfused blood delivered from a selected artery is clearly shown.
In
The measured perfusion-weighted data at different phase offsets (e.g., mi,n) is shown in (c), 606. Images (d), 608, and (e), 610, illustrate the estimated CBF-weighted and phase error maps, separately. By setting the estimated CBF signal level (e.g., image (d)) to 1.0, the mean absolute signal levels at each phase offset of the right ICA were observed to yield 0.79, 0.59, 0.28, 0.68, 0.65, 0.73, 0.61, and 0.41 in the order as shown in (c), thus illustrating the enhancement of the off-resonance corrected signal. As will be seen, the pattern of signal changes with different phase offsets is consistent with the simulation results. The SNR was improved by 47% by the proposed correction method compared to the signal obtained at 0° phase offset.
In the illustrated embodiments, the parameters used in simulation and in vivo experiments were: hamming-shaped RF pulses with 600 μs duration, 1.8 mm tagging slice thickness, gradient fraction 0.1, RF spacing 1500 μs, in-plane vessel-selective gradient amplitude 0.7 mT/m, gradient rotation rate of 11°, blood velocity 30 cm/s, tag duration 1.5 s, post-labeling delay 1 s with background suppression. A T2 of 275 ms and a T1 of 1680 ms was used in the simulation. One healthy subject was scanned in Toshiba® 3T Titan magnet, FFE2D readout (FA/TR/TE: 20 0/9/3.4 ms, matrix size 642, imaging slice thickness 10 mm, total TR 6 s, single slice). Three averages at each phase offset and eight offsets (n-1:8, −120 °, −90°, −30°, 0°, 30°, 60°, 90°,120°) were obtained, resulting in an acquisition time of around 4.5 minutes. The data acquired at the phase offset of −60° was discarded due to extensive motion artifact.
Thus, embodiments effectively restore the signal loss due to off-resonance artifact in vessel-selective pCASL and thus provide higher SNR compared to original VS-pCASL. Embodiments may also yield improved signal to noise ratio (SNR), for example, when compared to the signal without correction obtained, for example, at 0° phase offset.
Although the above embodiments were described primarily with respect to the VS-pCASL technique, the teachings herein are applicable for off-resonance correction of other territory-selective ASL techniques such as, for example, and without limitation, Ouyang et al., “Regional Perfusion Imaging Using pTILT,” Journal of Magnetic Resonance Imaging, doi: 10.1002/jmri.24346 (2013).
As demonstrated by both simulated and human results, the efficiency of vessel-selective pCASL labeling can be degraded in the presence of off-resonance effects. However, as demonstrated above, the proposed modified multiple-phase correction method can effectively restore signal loss due to off-resonance artifact and thus provide higher SNR in vessel-selective pCASL. Another benefit to application of the multiple phase correction method in single-artery pCASL is that, unlike the non-vessel-selective pCASL sequence, for the single-artery labeling, the signal model shown in
While certain embodiments have been described, these embodiments have been presented by way of example only and are not intended to limit the scope of the inventions. Indeed, the novel embodiments described herein may be embodied in a variety of other forms; furthermore, various omissions, substitutions and changes in the form of the embodiments described herein may be made without departing from the spirit of the inventions. The accompanying claims and their equivalents are intended to cover such forms or modifications as would fall within the scope and spirit of the inventions.
