Techniques which can be used to follow the position of invasive devices in real-time using magnetic resonance (MR) are described. Tracking of an invasive device is made possible by incorporating one or more small RF coils into the device. These coils detect MR signals from only those spins near the coil. Pulse sequences which employ nonselective RF pulses to excite all nuclear spins within the field-of-view are used. Readout magnetic field gradient pulses, typically applied along one of the primary axes of the imaging system, are then used to frequency encode the position of the receive coil(s). Data are Fourier transformed and one or more peaks located to determine the position of each receiver coil in the direction of the applied field gradient. Subsequent data collected on orthogonal axes permits the localization of the receiver coil in three dimensions. The process can be repeated rapidly and the position of each coil can be displayed in real-time.

Pulse sequences that permit selective detection of moving spins in a magnetic resonance image have been developed. Experiments were performed by the authors to produce projected angiographic data without the use of contrast agents, with the intensity of each image pixel determined by the macroscopic velocity of the detected spins. With this method, suppression of nonmoving spins is essentially complete, yielding a high dynamic range in signal intensity for detected vessels. Selective detection of moving spins is not dependent on pulsatile flow. Consequently, not only arterial structures, but also venous structures can easily be visualized. High-resolution angiographic images can be obtained by combining the flow experiment with surface coil techniques.

MR-guided interventional procedures can be performed with full patient access with use of an open-configuration, superconducting MR magnet with near real-time imaging and interactive image plane control.

Motion artifact suppression technique (MAST) for MR imaging. J. Comput. Assist. Tomogr. 11, 369-377 (1987). 11. Q. S. Xiang, M. J. Bronskill, R. M. Henkelman, Two-point interference method for suppression of ghost artifacts due to motion. J. Magn. Reson. Imaging 3, 900-906 (1993). 12. Q. S. Xiang, R. M. Henkelman, Dynamic image reconstruction: MR movies from motion ghosts. J. Magn. Reson. h a g-ing 2 , 679-685 (1992). 13. P. R. Moran, A flow velocity zeugmatographic interlace for NMR imaging in humans. Projection angiograms of blood labeled by adiabatic fast passage. Magn. Reson. Med. 3, 454-462 (1986). 18. W. Zhang, A quantitative analysis of alternated line scanning in k space and its application in MRI of regional tissue perfusion by arterial spin labeling. J. M a p. Reson. Series B 19. G. H. Glover, Phase-offset multiplanar (POMP) volume imaging: a new technique.

The improvement of MRI speed with parallel acquisition is ultimately an SNR-limited process. To offset acquisition-and reconstruction-related SNR losses, practical parallel imaging at high accelerations should include the use of a many-element array with a high intrinsic signal-to-noise ratio (SNR) and spatial-encoding capability, and an advantageous imaging paradigm. We present a 32-element receive-coil array and a volumetric paradigm that address the SNR challenge at high accelerations by maximally exploiting multidimensional acceleration in conjunction with noise averaging. Geometric details beyond an initial design concept for the array were determined with the guidance of simulations. Magnetic resonance imaging (MRI) conventionally relies entirely on magnetic field gradients to encode spatial information (1,2). Parallel MRI, as exemplified by the simultaneous acquisition of spatial harmonics (SMASH) (3) and sensitivity encoding (SENSE) (4) techniques, facilitates imaging by partially shifting the burden of spatial encoding from magnetic field gradients to parallel receive coils. One advantage of parallel MRI is that it can improve data acquisition speeds beyond what can be achieved with conventional nonparallel approaches, without imposing additional stress on the gradients. The practical implementations of parallel imaging have typically enabled as much as fourfold accelerations for a wide range of clinical applications.It is well understood that the upper limit of the attainable acceleration factor is equal to the number of parallel receive channels/coils on a scanner system. In practice, however, acceleration is additionally limited by the signalto-noise ratio (SNR) of the system. While the maximum number of parallel receive channels on general-purpose clinical scanners is currently in the vicinity of eight, and can be expected to increase in the near future, the system SNR represents a fundamental challenge and is increasingly limiting as acceleration advances.The SNR limitation on the practice of parallel imaging manifests itself in the well-known scaling by the square root of total acquisition time (5), as well as in a spatially varying noise amplification pattern (3,4) that intrinsically depends on both the k-space sampling pattern and the coil array geometry. As the acceleration factor increases, the noise amplification tends to be aggravated, which, combined with the decrease of noise averaging resulting from the reduction in total acquisition time, leads to an increasingly fast deterioration of image SNR. This property of parallel imaging SNR tends to project a somewhat pessimistic picture for the clinical use of high acceleration factors that exceed the current norm of 2-4. Nevertheless, in the push for higher accelerations, there are various strategies to which one may resort to mitigate SNR deterioration and manage image quality. One such strategy is to raise the baseline SNR by employing, e.g., high-fieldstrength systems, contrast-enhanced (CE) acquisitions, or high-SNR pulse sequences. Anothe...