COBRA - Continuously Ordered B0 Readout Acquisition

Martyn Paley¹, Steven Reynolds¹, Sarah Calvert², and Allan Pacey²

¹Immunity, Infection and Cardiovascular Disease, University of Sheffield, Sheffield, United Kingdom, ²Oncology and Metabolism, University of Sheffield, Sheffield, United Kingdom

Synopsis

A novel encoding method known as Continuously Ordered B0 Readout acquisition or COBRA is described. The method uses an additional B0 coil with a unique field at every point in space to perform Volume Frequency Encoding. The field increases as a monotonic function within the coil and can be sorted to the appropriate 3D location for reconstruction. 3D data sets have been acquired and reconstructed at 9.4T.

Purpose

To develop a fast imaging method using 3D volume frequency encoding (VFE).

Background

Use of auxiliary field coils to create steps which allow simultaneous parallel slice imaging such as in B1Ac-MAMBA (1,2) aim to increase MR throughput dramatically. A new simultaneous volume frequency encoding (VFE) imaging method known as COBRA (Continuously Ordered B0 Readout Acquisition) is introduced which uses a B0 field varying uniquely at each point within the 3D imaging volume. The frequency at each point can be mapped a priori and then used in conjunction with a single broadband pulse-acquire sequence to provide complete, rapid volume coverage in a single FID.

Theory and Methods

The B0 fields from a supplementary ten turn helical coil, with a variable pitch, set proportional to the square root of the z coordinate were calculated using the Biot-Savart law in Matlab (Mathworks, Natick, CA), figure 1, with a 29x29x29 matrix. Figure 1 shows the vector field from the simulated COBRA coil and Figure 2 shows the calculated Bz field map from positions ascending through the coil. Figure 3 shows the positive, sorted Bz field values from within the coil varying with a continuous, monotonic function. A prototype COBRA coil was wound on a 5mm, water filled NMR tube according to the above equation using 0.2 mm diameter wire and placed in a 10mm RF coil at the center of an Avance III scanner operating at 9.4T/400MHz (Bruker, Ettlingen, DE). A pulse-acquire sequence was collected with a receiver bandwidth of 100ppm and a RF pulse bandwidth of 10KHz with 24389 acquisition points. A current of 0.25A was used to generate a maximum additional COBRA field of 0.3mT as measured from the maximum frequency shift produced. An image volume set was created by reordering the frequency data to the known 3D locations as modeled for the COBRA coil a priori, using the sort index.

Results

Figure 4 shows the indices of the field sorting vector used to reconstruct the images by assigning signal to the predicted 3D field locations. Each 3D location has a unique B0 allowing Volume Frequency Encoding (VFE). Figures 5 shows a simulated, unsmoothed 1DFT of a COBRA FID calculated for a centrally located 14x14x29 rectangular phantom with increasing intensity and added Gaussian noise and Figure 6 shows the reconstructed image. Figure 7 shows the acquired 9.4T COBRA data set for a uniform cylindrical Gd-doped water phantom filling the constructed coil with similar characteristics to the simulation shown in Figure 5. Figure 8 shows the 3D sorted, positive field masked, 29x29x29 volumetric COBRA images reconstructed from a single FID.

Discussion and Conclusion

As only static field encoding is used, the acquisition is completely silent and encodes an entire volume within a single FID. The images have a nominal resolution of 0.22x0.22x0.44 mm and thus have relatively low SNR in a single shot. Improved encoding but with restricted physical access can be produced using conical coils. Helical coils could also be used together with a static Z-gradient. Fiducial markers might improve registration of data sets to field calculations. Further development of the COBRA 3D volume frequency encoding (VFE) method could be a useful addition for rapid 3D MR imaging particularly in cases where peripheral nerve stimulation or acoustic noise is a concern.

Acknowledgements

We acknowledge funding from MRC grant MR/M010473/1

References

1. Feinberg DA, Setsompomp K, Ultra-fast MRI of the human brain with simultaneous multi-slice imaging.J Magn Reson. 2013; 229: 90–100

2. Paley M et al., B1AC-MAMBA: B1 array combined with multiple-acquisition micro B0 array parallel magnetic resonance imaging. Magnetic Resonance in Medicine 2003; 249:1196–1200.

Figures

Figure 1. Calculated field of 10 turn COBRA coil on 29x29x29 matrix

Figure 2. Map of field through 29 planes ascending through the COBRA coil

Figure 3. Sorted positive fields from the COBRA coil showing monotonic increase

Figure 4. The sort vector used to reorder the field values to 3D location

Figure 5. Simulated COBRA data set for a rectangular 14x14x29 pixel phantom.

Figure 6. Simulated image for a rectangular 14x14x29 phantom with added Gaussian noise.

Figure 7. COBRA data set acquired at 9.4T from a uniform cylindrical water phantom

Figure 8. Image reconstructed on 29x29x29 matrix from the data set shown in Figure 7 using the sort vector shown in Figure 4.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)

3174