ABSTRACT
Recently, we introduced a new numerical approach to the design and optimization of NMR selective pulses, which we have christened "SPINCALC" (J. T. Ngo and P. G. Morris, Biochem. Soc. Trans. 14, 1271 (1986); J. T. Ngo and P. G. Morris, Magn. Reson. Med. 5, 217 (1987]. The first practical application of pulses generated by SPINCALC is demonstrated on a standard 0.5-T clinical MRI system. Results are shown for single phase pi pulses suitable both for selective inversion and for selective refocusing. The extension of SPINCALC to multidimensional pulses is illustrated by the design of a two-dimensional pi pulse.
Subject(s)
Algorithms , Magnetic Resonance Imaging , Magnetic Resonance Spectroscopy , Fourier Analysis , Humans , Models, Structural , Models, TheoreticalABSTRACT
The need for NMR selective pulses in magnetic resonance imaging and spectroscopy is reviewed. The shortcomings of the current generation of pulses are discussed and the need for new categories of pulse identified. Strategies for selective pulse design are outlined and two numerical optimization methods, simulated annealing and SPINCALC (a method recently introduced by us: J. T. Ngo and P. G. Morris, Magn. Reson. Med. 5, 217 (1987], are discussed in detail. Their use is illustrated and compared for the design of pi/2 phase-compensated pulses. Both methods require substantial amounts of CPU time, with simulated annealing the more demanding. Unconstrained, simulated annealing also tends to produce pulses with discontinuous waveforms. A crude two-dimensional pulse derived from a low flip angle approximation is illustrated.
