Novel techniques for 7 tesla breast MRI

Abstract

This thesis introduced several new techniques to the field of 7 tesla breast MRI, enabling high field multi-parametric MR imaging and, potentially, patient specific treatment planning. Chapter 2 described the development of a RF coil setup for bilateral breast MR imaging and 31P spectroscopy. This setup was constructed in three different modules. The first module consists of the transmit/receive part of the RF coil. Two quadrature resonators were positioned next to each other, both tuned at the frequencies of 1H and 31P. Normally, coils placed close to each other can be decoupled by partial overlap to minimize mutual inductance. However, it is more complicated for double-tuned resonators as the amount of overlap is frequency dependent. The addition of a novel floating loop does provide good decoupling of both transmitters for two different frequencies. The second module is a high-density receiver array, of 26 receiver loops, tuned for the 1H frequency. With this receiver array, scan times have been accelerated up to a factor of 8 by using parallel imaging techniques. The final module consists of an interface to the scanner hardware. This module enables the use of parallel transmit techniques for the 1H frequency by connecting each resonator to an individual RF amplifier. In addition, it makes parallel imaging techniques possible for the 31P frequency, by connecting each resonator to a separate receive path. With this coil, high resolution dynamic contrast enhanced (DCE) images (0.7 mm isotropic) can be acquired in 90 seconds. At these resolutions, subtle motion can result in quite severe errors in the calculation of enhancement curves, potentially misclassifying tumours. Therefore, the use of fat suppression is essential. In chapter 3 we investigated what fat suppression technique results in the best image quality at 7 tesla. In this study two potential candidates, Dixon and water selective excitation (WSE) were investigated and images were scored quantitatively and qualitatively. Whereas it is already known that Dixon results in very good images for DCE MRI at 3 tesla, it was unknown which method was the best choice for 7 T. Our results show that although the fat suppression of Dixon was slightly better, small anatomical details were lost when compared to WSE. Altogether, the overall image quality of WSE was better than Dixon. To further improve on the fat suppression of the WSE method, the full potential of the used RF coil was investigated in chapter 4. The multi-transmit capabilities of the RF coil makes it possible to transmit the RF pulse on each channel with different resonant frequencies. This means, for the WSE pulse to perform well, a B0 offset between the left and right breast is acceptable. With this method, named frequency shimming, an improvement in fat suppression was found for a group of subjects, without compromising on image quality. As there is no penalty in terms of image quality for using this method, and the workflow of the technician can remain unchanged, implementation in the clinical routine is easy to achieve. Chapter 5 describes a technique to, ultimately, increase the resolution of EPI sequence with minimal geometrical distortions by using an unshielded gradient insert coil. The use of gradient insert coils is not novel as such, however, we suggest to minimize the eddy current effects by using such a coil in conjunction with the whole-body gradient system. Omitting the shielding of a gradient coil results in eddy currents in the cryostat. The effects of the eddy currents, often resulting in severe reduction of image quality, could be counteracted by driving the whole-body gradient system at the same time. Although this technique was demonstrated with a coil designed for breast imaging at 7 T, a similar approach is valid for gradient insert coils designed for other body parts, such as the head, but also applies to lower field strengths.