Parallel Imaging in Clinical MR Applications

18 High-Resolution Imaging of the Brain (p. 183-184)

Roland Bammer and Scott Nagle

CONTENTS
18.1 Introduction 183
18.2 Structural MRI 185
18.3 MR Angiography 190
18.4 Quantitative/Functional MRI 193
18.5 Pediatric MRI 196
18.6 Conclusion 196
References 197

18.1 Introduction

Resolution enhancement in MRI is of great potential for increased diagnostic accuracy in brain and spine imaging. With the advent of high-. eld systems increased signal-to-noise ratio (SNR) affords smaller voxel sizes, but overall scan time is still a limiting factor for high-resolution brain imaging in a clinical setting. Parallel imaging is of great bene. t since it allows us to achieve high-resolution 2D and 3D acquisitions in clinically acceptable time frames. In addition, parallel imaging can also diminish the amount of image blurring and geometric distortions leading to obvious resolution and quality improvements without altering the acquisition matrix size.

This chapter critically addresses the general advantages and limitations of high-resolution neuroimaging in concert with the additional capacity provided by parallel imaging. Specifically, the role of parallel imaging in high-resolution structural MRI, magnetic resonance angiography, and functional MRI in the broader sense (i.e., diffusion and perfusion MRI as well as classical functional MRI) are discussed. The improved scanning efficiency of parallel imaging methods can be applied in a number of fruitful ways in the . eld of brain MR imaging. As described in the . rst part this book, these parallel-imaging strategies cleverly incorporate the spatially varying sensitivity pro. les of multiple-channel receive coils in order to reduce the number of k-space measurements necessary to reconstruct an image (Hutchinson and Raff 1988, Kwiat et al. 1991, Sodickson and Manning 1977, Pruessmann et al. 1999). In conventional Cartesian k-space sampling schemes, this is typically realized by decreasing the number of phaseencoded steps. In cases of high SNR, the reduction of phase-encoded steps by a factor R (accompanied by its obligatory R decrease in SNR Pruessman et al. 1999) can be used to increase either spatial resolution or temporal resolution. These advantages are not mutually exclusive and it depends on the specific scan protocol whether one favors more rapid scanning or higher resolution. In addition, the artefact and blurring associated with several multiple-echo or long-readout sequences can be mitigated by parallel- imaging techniques (cf. Chap. 10).

Increasing the spatial resolution may allow the detection of smaller lesions, may better characterize the internal structure of larger lesions (e.g., calcification, blood products, demyelination, cystic components, etc.), and may better delineate the lesion boundaries with respect to normal anatomy, improving the accurate localization of a lesion (especially important in the prepontine, suprasellar, cerebellopontine angle, cavernous sinus, orbital, and Meckel’s cave regions). The imaging of white matter disease, stroke, neoplasm, and vascular disease could all ben e. t from these advantages. Pediatric and neonatal brain imaging also demands fast, high-resolution imaging because of the relatively small brain size and the dif. culties in keeping a child still throughout the scan. Similar considerations apply also for imaging the spinal cord. Three-dimensional spoiled gradient- echo sequences, used in the evaluation of mesial temporal sclerosis in the work-up of seizures, tumor treatment planning, and voxel-based morphometry in neurodegenerative disorders, could benefit from the use of parallel imaging in both phase-encoded directions to further increase resolution without increasing scan time (Weiger et al. 2002a).

Conversely, shortened scan times alone can increase patient throughput, resulting in obvious operational and patient comfort improvements. Simply decreasing scan time reduces the risk of patient motion degrading a study. A number of other creative methods for further reducing motion artefacts through the use of parallel imaging have been proposed and demonstrated (Bammer et al. 2004, Kuhara and Ishihara 2000, Bydder et al. 2002, 2003, Atkinson et al. 2004).