Version française
Institutes/ Institute of biomedical imaging (I²BM)/ NeuroSpin (D. Le Bihan)/ Imaging and spectroscopy laboratory (LRMN)/ Research Programs

Imaging and spectroscopy laboratory - Research Programs

 Tackling B1 inhomogeneity using strongly modulated RF pulses
A new pulse technique for counteracting RF inhomogeneity at high fields has been developed. The method, inspired from nuclear magnetic resonance quantum computing [1], makes use of the detailed knowledge of the voxels' B1 and DB0 amplitude 2D histogram to generate, through an optimization procedure, gates where the flip angle is made uniform. The use of such 2D histogram instead of the parameters’ joint spatial distribution decreases substantially the complexity of the problem, allowing an optimization algorithm to find an RF pulse solution in less than 30 sec. In addition, the procedure is based on an exact calculation and does not use any linear approximation.
We’ve been using the sequences reported in [2, 3] to measure B1 and DB0 over some volunteers’ brains using a standard square pulse. Results at 3T from this measurement scheme are displayed in figure 1, showing B0 map, B1 map and 2D histogram. The 3D data is used to compute a {B0, B1} bi-dimensional histogram. The performance of a candidate pulse can thus be calculated quickly by looping over a few tens of values. An optimization procedure combining a genetic algorithm with a simplex method then takes around 30 sec to return a waveform, no matter what the nominal flip angle is. Once the new strongly modulating pulses are available, they can be inserted into new sequences and tested.
 
 

 Fig. 1. Measurement of DB0 and B1 (using a standard square pulse) in one volunteer’s
brains at 3T. a: B0 map,  b: B1 map, c: 2D histogram.
 
 

Figure 2 shows the measurement result of the sine of the flip angle over a full brain at 3T when using a standard square pulse and a strongly modulating pulse with nominal flip angle of 90°. The standard deviation of the sine of the flip angle over the whole brain is hereby reduced by a factor of 15. Additional calculations show that BIR4 adiabatic pulses [4] seem to require 2 to 3 (up to 8 at 7 T) times more energy to reach the strongly modulating pulses’ performance.
 
 
  
Fig. 2. Measurement result of the sine of the flip angle over a full brain at 3T
when using a standard square pulse (top row) and a strongly modulating pulse (bottom row) with nominal flip angle of 90°.
 
An even bigger challenge exists at 7T. At that field strength, the ratio of the maximum to the minimum of transmitted B1 can reach a factor 4 or 5 (while it reaches a factor of approximately 2 at 3T) with a quadrature head coil. We have tested the strongly modulating pulses at that field. For a 90° pulse, we have measured a factor of 3 of reduction of the standard deviation of the sine of the flip angle, although this number was mostly penalized by the regions where large B0 inhomogeneity were present and where B1 was particularly low. A 12° pulse was also implemented in a 3D GRE sequence with TR = 100 ms, TE = 4 ms, resolution = 1.4 x 1.2 x 1.4 mm3. Figure 3 shows coronal slices extracted from these 3D data sets, corresponding to each pulse scenario (square pulse and strongly modulating pulse), along with a 1D profile going through the centre. The gain is three-fold: 1) the bright spot in the centre has been removed, 2) a very large amount of signal in the lower region of the brain has been recovered and 3) some signal has also been gained in the periphery of the brain. Overall, over the brain, the signal has become much more uniform.
References: [1] E. M. Fortunato et al. J. Chem. Phys. 116:7599-7606 (2002). [2] V. L. Yarnykh. MRM 57:192-200 (2007). [3] A. Amadon et al. ISMRM 2008. [4] Bernstein et al. Handbook of MRI Pulse Sequences. Elsevier Academic Press (2004).
 
 

Fig. 3. Coronal slices extracted from 3D gradient echo images at 7T using different excitation pusles: a) square pulse, b) strongly modulating pulse. A 1D profile going through the center is shown on top of the figure.
Real time processing of MRI
Molecular and metabolic MRI for diagnosis of brain diseases
High angular resolution diffusion MRI
New RF coil desing