Altered phase-encoding order for reduced sensitivity to motion in three-dimensional MR imaging.

A method of reordering phase and slab encoding that can be used to address some of the inherent problems due to motion in three dimensional imaging is described and implemented. The method is shown to be more robust with respect to reducing artifacts resulting from several fundamental types of motion. It can be readily implemented on a standard magnetic resonance imager with essentially no increase in total imaging time. Results of simulations and phantom and in vivo experiments are presented.     

Three-dimensional motion tracking with volumetric phase contrast MR velocity imaging

Motion tracking based on single-slice cine-phase contrast magnetic resonance imaging data has limitations. In the presence of nontrivial three-dimensional motion and deformation, volumetric data are necessary for accurate reconstruction of material point trajectories. A three-dimensional Fourier tracking method that uses volumetric data for motion tracking is presented. The method reconstructs a material point trajectory by computing its various harmonics. For any given temporal sampling rate, a frequency domain perspective of the tracking problem indicates that the method is accurate in estimating all reconstructible harmonics of a trajectory. The algorithm incorporates an intra-voxel linear spatial model into the integration to address potential tracking performance degradation due to possibly reduced spatial resolution, which may be most relevant in the slice direction (z) if the volumetric data are obtained as multiple two-dimensional slices. The tracking method was evaluated on computer-generated data sets that simulated various motion patterns. The method was also tested with two sets of in vitro data obtained using a phantom, one acquired as multiple two-dimensional slices and the other using a three-dimensional sequence capable of higher spatial resolution in the z direction. These studies demonstrated that the algorithm can achieve high sub-voxel tracking accuracy.     

In-plane motion correction for MR spectroscopic imaging

A motion-detection method is described that is specifically suited for MR spectroscopic imaging (MRSI) studies. Information on in-plane rotation and translation of the subject was obtained using external spatial reference markers that are uniquely identified via their chemical shift. The marker locations were obtained directly from the acquired data at each encoding step, and no additional data acquisition was required. This method was applied to brain ¹H MRSI studies that include subcutaneous lipid signals, which otherwise result in enhanced sensitivity to subject motion.     

Development and evaluation of tracking algorithms for cardiac wall motion analysis using phase velocity MR imaging

Phase velocity magnetic resonance imaging (MRI) has shown considerable potential for tracking distinct regions of the myocardium throughout the cardiac cycle. Phase contrast MR imaging produces multiple images, each phase encoded for velocity in a different direction, in which individual pixels depict the local motion of the tissue. In this work we present in detail three algorithms for tracking motion based on these images. Both simulated and phantom data are used to examine some of the problems encountered in practice in tracking points based on velocity maps. Solutions to these problems are offered when possible. The impact of noise and low order phase errors in the data on each of the three tracking approaches is examined. In addition, problems due to tissue expansion and contraction, to 2D versus 3D tracking, and to round off errors from motion which is small relative to pixel size or slice thickness, are considered. An example using data obtained in vivo is included to demonstrate the efficacy of the best of the three tracking algorithms in measuring left ventricular circumferential shortening preinfarct and postinfarct in a canine model.     

Compensation for effects of linear motion in MR imaging

Various compensation methods for different types of motion during MR image acquisition have been proposed. Presented here is a postprocessing scheme for eliminating artifacts due to linear, intra-slice motion of known velocity. The data for each phase encoding or "view" acquired from a moving object are shown to differ from those which would be measured from the stationary object by a phase factor which depends on the object's displacement from a reference point. This derivation is then used to propose a correction scheme for linear motion in which all phase encodings measured at different positions of the moving object are collapsed onto the same reference position. After subsequent reconstruction, the object appears perfectly "focused." By selection of different reference positions, the method permits positioning of the compensated object as desired within the field of view of the image. This property allows the method to be extended to create sequences of corrected images with smooth object motion between frames of the sequence. The basic correction scheme and its variations were tested experimentally in phantom studies with velocities as large as 8 cm/s.     

Validation of cine phase-contrast MR imaging for motion analysis

The accuracy of cine phase-contrast magnetic resonance (MR) imaging for motion analysis was evaluated. By using a rotating phantom and postprocessing algorithm for phase tracking, errors arising during data acquisition were identified and compensation methods were developed. A spatially varying background phase offset in the velocity images was found to be due to eddy current-induced fields. The magnitude of the offset was in the range of 0–20 cm/sec, which is of the same order of magnitude as cardiac contractile velocities. Background offset is thus an important source of error in tracking cardiac motion. Study of different tracking algorithms revealed the need for an integration scheme using motion terms higher than velocity. Also, considerable improvement in the accuracy and stability of the predicted trajectories was obtained by averaging the trajectories proceeding both forward and backward in time from the starting point. With the algorithm developed, the motion of the phantom was tracked through a complete rotation of the phantom to an accuracy of 2 pixels.     

Overcoming motion in abdominal MR imaging

Anatomic structures that move periodically during the acquisition of data for an MR image become multiple ghosts in the phase-encoding direction. There is a constant spacing in pixels between consecutive ghosts, which is equal to the number of cycles of motion that occurred during the acquisition of data. The intensity of ghosts depends on the intensity of the moving structure and the number of pixels over which the motion occurred. No single method is completely satisfactory at suppressing motion artifacts. The major attributes and limitations of each method are summarized in Table 2, with plus (+) signs denoting merit. Theoretically, some methods perform better in reducing the intensity of ghosts and restoring the image intensity to its proper place. This certainly is not the final criterion, however. Some methods reduce the blurring in addition to suppressing the ghosts, or they suppress ghosts without prolonging the time for imaging. Certain methods also reduce ghosts from other kinds of motion. It is very appealing for a method to function without monitoring. The success of monitoring often depends too much on the cooperation of both the patient and technologist. The theoretical performance, attributes, and deficiencies of the various methods have been combined into a subjective overall rating in the last column of Table 2. All of the methods can be effective under the appropriate circumstances. Moreover, the methods are not mutually exclusive. It is advantageous, therefore, to combine methods to achieve even greater suppression. For example, physical restraint can be used for all but the most uncooperative patients. Most imaging techniques can be designed with gradients that rephase the signals from moving structures. Then other methods, such as averaging or reordering, can be applied as necessary. Fortunately, there are effective motion artifact suppression methods, even though not all are widely available yet on commercial equipment. Consistent suppression of motion artifacts will enhance the quality of MR images. Elimination of motion artifacts will improve the capability of MR to detect lesions and will provide a higher standard of performance for MR in the body.     

Method for measuring three-dimensional motion with tagged MR imaging

Recent methods of magnetic resonance imaging involve the placement of a grid of planes of saturation over the imaging plane; distortion of the grid corresponds to tissue displacement in two dimensions. An extension to this method that allows measurement of motion in the third dimension involves a second acquisition that tilts the grid, allowing analysis of motion normal to the imaging plane. A rotating phantom was used to verify the accuracy of the motion measurements, and the technique was applied to the heart wall and skeletal muscle. Phantom results show that the measure of z motion can be as accurate as that of x and y motion. Three-dimensional displacements of heart-wall and skeletal muscle are shown. With an accurate measure of three-dimensional motion, more complete analysis of heart-wall motion and contraction is possible.     

MR imaging, flow and motion.

The present work is intended as a nonmathematical review of the role of flow and motion in nuclear magnetic resonance (MR) imaging. A historical review of MR flow measurement techniques is given, followed by a short overview of flow models in vitro and in vivo. The theory behind the influence of motion on the modulus and phase MR signal information is discussed and effects such as washin/washout, flow-induced signal void, phase offset, and phase dispersion are defined. A simple approach to the concept of MR angiography is given, and methods for quantitative flow measurements such as the phase mapping technique, are surveyed. Aspects of the measurement of diffusion and microcirculation are given, and finally, an overview of the role of MR flow imaging in present and future clinical application is given.     

MR velocity imaging by phase display

The ability of the nuclear magnetic resonance signal to encode information about macroscopic motion has been recognized since the works of Hahn and Carr and Purcell. In the medical imaging setting this ability has led to a variety of ingenious magnetic resonance flow imaging schemes that ultimately may become competitive with X-ray angiography in sensitivity and specificity while remaining radically noninvasive. This work demonstrates that conventional spin-echo Fourier transform image acquisitions naturally encode a component of flow velocity that lies within the image plane. By displacing just the real part of the complex image data (phase display), the velocity distribution within the subject is revealed in image form. This method of flow imaging requires neither special pulse sequences nor image reconstruction and format software for its implementation. Further, images that intersect a flow channel longitudinally, demonstrating in-plane flow, yield an unusually large quantity of physiologic information per image. Phantom and in vivo flow images are presented. Also described is a phantom based on a rotating disk that enables calibration of the velocity/phase-shift constant for an untested pulse sequence.     

Respiratory ordered phase encoding (ROPE): a method for reducing respiratory motion artefacts in MR imaging

A method of reducing respiratory artefact when using the spin warp technique of magnetic resonance imaging is described. A respiratory signal is used to determine the order in which the rows of the data matrix are measured. The aim is to make the respiratory signal at the time of each phase encoding gradient a slowly varying function of the time integral of the gradient and hence of the row number. Data are collected during each cycle of the imaging sequence; thus the scanning time is not increased.     

Patient motion correction for multiplanar, multi-breath-hold cardiac cine MR imaging

PURPOSE: To correct for spatial misregistration of multi-breath-hold short-axis (SA), two-chamber (2CH), and four-chamber (4CH) cine cardiac MR (CMR) images caused by respiratory and patient motion. MATERIALS AND METHODS: Twenty CMR studies from consecutive patients with separate breath-hold 2CH, 4CH, and SA 20-phase cine images were considered. We automatically registered the 2CH, 4CH, and SA images in three dimensions by minimizing the cost function derived from plane intersections for all cine phases. The automatic alignment was compared with manual alignment by two observers. RESULTS: The processing time for the proposed method was <20 seconds, compared to 14-24 minutes for the manual correction. The initial plane displacement identified by the observers was 2.8 +/- 1.8 mm (maximum = 14 mm). A displacement of >/=5 mm was identified in 15 of 20 studies. The registration accuracy (defined as the difference between the automatic parameters and those obtained by visual registration) was 1.0 +/- 0.9 mm, 1.1 +/- 1.0 mm, 1.1 +/- 1.2 mm, and 2.0 +/- 1.8 mm for 2CH-4CH alignment and SA alignment in the mid, basal, and apical regions, respectively. The algorithm variability was higher in the apex (2.0 +/- 1.9 mm) than in the mid (1.4 +/- 1.4 mm) or basal (1.2 +/- 1.2 mm) regions (ANOVA, P < 0.05). CONCLUSION: An automated preprocessing algorithm can reduce spatial misregistration between multiple CMR images acquired at different breath-holds and plane orientations.     

Arterial input functions from MR phase imaging

A method is presented for obtaining high-sensitivity arterial input functions following bolus intravenous contrast agent administration. Arterial contrast agent is monitored by phase reconstruction of single-shot echo-planar images. During bolus injections of a gadolinium (Gd) agent in a baboon, data were acquired at the mid-abdominal aorta, and magnitude and phase-shift images were reconstructed. Pair-wise image subtraction was used to minimize phase aliasing. The phase-based method is shown to have a significant potential improvement in sensitivity compared to the magnitude approach. The phase method also has a general linear response to concentration. This method may have potential utility in quantitative imaging of blood flow and contrast agent kinetics.     

MR imaging of motion with spatial modulation of magnetization

A novel magnetic resonance imaging technique provides direct imaging of motion by spatially modulating the degree of magnetization prior to imaging. The preimaging pulse sequence consists of a radio-frequency (RF) pulse to produce transverse magnetization, a magnetic field gradient to "wrap" the phase along the direction of the gradient, and a second RF pulse to mix the modulated transverse magnetization with the longitudinal magnetization. The resulting images show periodic stripes due to the modulation. Motion between the time of striping and image formation is directly demonstrated as a corresponding displacement of the stripes. This technique can be used to study heart wall motion, to distinguish slowly moving blood from thrombus, and to study the flow of blood and cerebrospinal fluid.     

Ineffectiveness of averaging for reducing motion artifacts in half-Fourier MR imaging.

Two data sets for half-Fourier imaging (HFI) can be collected in the same time as one data set for conventional full Fourier imaging (FFI). The hypothesis is that averaging twice as much data in HFI does not make ghost artifacts caused by motion have less signal intensity than in FFI. This hypothesis was tested with images of a human subject by measuring the standard deviation within regions of interest containing ghosts. The control experiment involved measuring the standard deviation on images from the same data reconstructed with FFI. The images were formed after averaging of one to eight data sets from a collection of nine data sets acquired sequentially. Background ghosts or those in other regions of low intensity were less intense on images from HFI after twice as much averaging as in FFI, but this was not the case for ghosts superimposed on anatomic structures. This observation is explained by showing that an image obtained by means of FFI can be expressed in terms of two images obtained by means of HFI applied to the top and bottom halves of the data. The use of HFI to allow twice as much averaging without prolonging data acquisition time is not advantageous for reducing ghost artifacts caused by motion.     

Aortic dissection: sensitivity and specificity of MR imaging

Gated transverse magnetic resonance (MR) images of 54 patients (35 male, 19 female; aged 16-90 years) with suspected or known aortic dissection were reviewed by three cardiac radiologists without knowledge of clinical details. The reviewers independently determined the presence or absence and the type of aortic dissection. A confidence level was assigned for each diagnosis, and receiver operating characteristic curves were generated. The reviewer with extensive MR experience correctly identified 96% of the proved aortic dissections and all of the normal cases; the reviewer with moderate experience identified 96% and 84%, respectively; and the reviewer with minimal experience, 78% and 94%. The sensitivity at a specificity level of 90% was determined for each reviewer (100%, 96%, and 83%, respectively). MR imaging is highly sensitive and specific in the diagnosis of aortic dissection but does require considerable experience because of the need to recognize flow artifacts.     

Cardiac-gated phase MR imaging of aqueductal CSF flow

The direction of CSF flow within the cerebral aqueduct was studied by cardiac-gated magnetic resonance (MR) phase images in five healthy volunteers and 10 patients with presumably normal cerebral CSF circulation. Caudal CSF flow was observed during systole and cranial flow during diastole. Using phantom based calibrations of the imager, aqueductal CSF velocities of 3-5 mm/s were calculated. Cardiac-gated phase MR is a potentially major tool for the investigation of the CSF circulation.