Logo of the Physikalisch-Technische Bundesanstalt

Magnetic resonance imaging at ultrahigh fields: problems and potential solutions


In January 2009 the new 7-tesla whole-body MRI scanner of the Berlin ultrahigh field facility at the Max-Delbrück-Centrum (MDC) für Molekulare Medizin in Berlin-Buch was handed over to the users. The facility, a joint venture of MDC, Charité – Universitätsmedizin Berlin, Leibniz-Institut für Molekulare Pharmakologie (FMP) and PTB, is dedicated to translational research, i.e. the transfer of research results "from bench to bedside". As far as MRI is concerned, this ambitious goal requires the highest sensitivity and thus the highest field strength possible. Right now, this is represented by the new generation of ultrahigh-field (UHF) scanners, a term denoting whole-body magnets of at least seven tesla. The high sensitivity comes with a price tag, however. Radio-frequency (RF) related image artefacts, which at 7 T are prevailing in a form not known at lower field strengths, are massively restricting the practical usefulness of UHF scanners so far. To study and to understand these artefacts and to find means to avoid them, these were primary motives for PTB to engage into this project in the first place. First experiments were performed using an 8-channel transmit/receive head coil (Rapid Biomedical, Rimpar) and a cylindrical head phantom (i.d. = 20 cm) filled with agarose gel (Fig. 1). The image obtained when driving the coil in the conventional, circularly polarised (CP) mode is shown in Fig. 2. The very same mode at 1.5 T (corresponding to a 1H Larmor frequency of 64 MHz) would have produced a perfectly uniform image of the object. The 300-MHz image obtained at 7 T, however, with its pronounced annular signal void is almost useless for practical applications. In Fig. 3 we depicted the relative amplitude maps of the RF magnetic fields produced by each coil element individually. These maps, obtained by special MR imaging protocols, indicate the signal void of Fig. 2 is not produced by the individual elements. It is rather a consequence of destructive interference when the eight vector fields are superimposed. This is good news, however, as it implies that the homogeneity of the image can indeed be substantially improved, if we abandon the CP mode and optimize the way we drive the coil. Thanks to the multichannel transmit array the Berlin 7 T (as the first UHF scanner in Germany) is equipped with we do actually have the technical means to pursue this strategy [1].

Figures 4 and 5 show results even exceeding this goal. In both cases the "Transmit SENSE" technique [2, 3] has been employed applying the spatial encoding principles normally used for signal detection only already in the excitation phase of the experiment. Thus it becomes possible to "write" arbitrary patterns into the object (and re-read them later on). Using the identical coil and phantom as before and displaying the results on the same length scale, Figs. 4 and 5 show Transmit SENSE images obtained with different target patterns. While the checkerboard (Fig. 4) is just a reference pattern, the flat square in Fig. 5 already represents a meaningful application: homogeneous excitation within and only within a predefined target volume. In the depicted image the signal intensity inside the square is varying by about 5 – 10% while all unwanted signals from outside are suppressed by at least a factor of 50. In summary it can be said that already the very first experiments with the new system reproduced the severe RF issues as they were expected from previous computer simulations. But even more importantly it could also be demonstrated that parallel excitation in general and Transmit SENSE in particular are suitable techniques to overcome these problems.

[1] Seifert et al., J Magn Reson Imag 26 (2007) 1315
[2] Katscher et al., Magn Reson Med 49 (2003) 144
[3] Ullmann et al., Magn Reson Med 54 (2005) 994


[Translate to English:] « Zurück