Analysis and manipulation of field inhomogeneity effects in magnetic resonance imaging

Abstract

In the thesis the effects of macroscopic field inhomogeneities in MR imaging are analyzed and manipulation. Field inhomogeneities disturb the signal phase and invoke image distortions. Phase images are shown to serve as an effective means to generate and manipulate image contrast. In chapter 2 the phase derivative is shown to provide positive contrast and allows paramagnetic and diamagnetic field perturbing structures to be discriminated. Phase derivative magnitude images provide positive contrast surrounding field perturbing objects, but are also sensitive to field gradients. In chapter 3, the specificity is shown to be increased by calculating the Laplacian of the signal phase, since it is zero in regions of constant or linearly varying susceptibility and nonzero at abrupt changes in susceptibility, for instance, at a single point, a ridge, an interface, an edge or a boundary. In chapter 4, the discriminative property of the signal phase is shown to allow detection of micro-calcifications in ex vivo breast specimens. In chapter 5, the phase derivative is shown to not only provide contrast, but also to be applicable to reduce signal loss due to signal dephasing induced by macroscopic field inhomogeneities. This macroscopic signal loss induces R2 signal decay, but does not reflect a tissue property. In chapter 5 it is shown that this unwanted effect of a field distortion on the signal can be reduced by dividing the acquired signal by an estimate of the signal dephasing, which is obtained by integrating the phase derivative over a voxel. The second part of the thesis focuses on image distortions induced by field inhomogeneities. In previous work the center-out Radial Sampling in Off-Resonance (co-RASOR) technique was shown to accurately localize a field perturbing object. Co-RASOR applies a global frequency offset to center-out acquired data to focus signal pile-up onto the exact geometrical center of a field perturbing object. In chapter 6, the required frequency offset is shown to be more efficiently applied during reconstruction than during signal acquisition. Reconstruction co-RASOR is furthermore demonstrated to be more flexible than acquisition co-RASOR and to allow the center of an object to be located automatically. In chapter 7, the acquisition duration of the 3D co-RASOR is addressed. It is shown that - under certain conditions - two 2D co-RASOR acquisitions, instead of one 3D acquisition, allow the acquisition time to be reduced from several minutes for the 3D acquisition to less than 4 seconds for the dual-plane technique, while maintaining the 3D accuracy. By applying the dual plane co-RASOR acquisition a HDR brachytherapy source could be tracked with 1 mm accuracy with a frame rate of approximately 0.25 Hz. In chapter 8, the pros and cons of single point imaging (SPI) as a solution for MR imaging artifacts in the presence of field inhomogeneities are discussed. SPI is attractive in that it applies phase encoding in all three dimensions whereby it becomes insensitive to image distortions induced by field inhomogeneities. Data acquired with SPI is compared to data encoded by applying frequency encoding along one encoding direction. In addition, the potential of SPI data is illustrated by phantom experiments. Furthermore, promising methods to decrease the acquisition duration are outlined.