Volume 10, No. 2, 1999, while the articles comprising Part II are published herein. In order for this collection of articles to be of greatest utility, contributors were encouraged to present their new results in a broader context. The authors succeeded admirably in this enterprise, providing lucid background material for those members of the IJIST readership who are proficient in image analysis but less so in MRI.
A striking example of the interplay between MRI acquisition methodology and more traditional signal processing is provided by the work on a new imaging method called PERL in a paper by Rybicki, Patz, Hrovat, and Pulyer and a companion paper by Patz, Hrovat, Pulyer, and Rybicki. These authors introduce a novel MR imaging procedure utilizing a magnetic field geometry which is periodic in one dimension and linear in the other. In this way, it is a significant departure from traditional imaging, and requires both specialized rf coil geometry and an unconventional reconstruction algorithm. Aside from the specific capabilities of the PERL technique, this work should provide stimulus for the further development of nonstandard magnetic field geometries in MRI.
An important advance in spatial selectivity was the recent development of the CARVE technique by Sers หa and Macura. In this method, arbitrarily complex regions of excitation may be created. This permits precise tailoring of signal acquisition to the region of interest. In addition to this high degree of flexibility, CARVE sequence design can be optimized around a variety of parameters, permitting, for example, the demands on the gradient system to be held to a minimum. While the original work on CARVE was restricted to two dimensions, a three-dimensional extension of the method has recently been developed. This allows for volume excitation which corresponds to the shape and size of realistic samples. In the present work, the authors extend and generalize their previous analyses to account for the off-resonance and relaxation effects which play a role in virtually all realistic studies.
Image registration issues have long played an important part in MRI studies. One application is the registration of two-dimensional MRI images with other planar modalities, such as computerized tomography or positron emission tomography. Specific issues arise when sequential MRI images are obtained. In the paper by Alexander, a new algorithm is presented for registration of such images. The algorithm is highly efficient and insensitive to noise. The framework developed is very general, and permits, for example, immediate extension to three-dimensional data sets.
Standard MRI reconstruction methods such as those based on simple two dimensional Fourier transformation of k-space data or projection reconstruction have the advantage that they are fully