Quantitation of various NMR properties of tissues elucidates information about the state of the underlying tissues. We would like to quantitatively measure T2* to attempt to understand the mechanisms underlying the BOLD effect and to tune and optimize T2* weighted fMRI sequences. Certain regions of the brain suffer from susceptibility artifacts from the nasal sinuses and ear canals. Certain neuro-biologically interesting regions of the brain are difficult, if not impossible to image with BOLD fMRI because of the "susceptibility holes" in pre-frontal cortex and along the inferior side of the temporal lobes.

Although echo planar imaging remains the work horse in the neuroscience community, other fast imaging techniques, like spiral, are frequently employed. Spiral k-space trajectories have been used to collect data in a variety of applications for a number of years.^1,2 It was quickly noticed that spiral k-space trajectories resulted in localized blurring of the images in the presence of B₀ inhomogeneities. Noll^3,4 et al. found ways to correct this blurring with a number of method including multi-frequency reconstructing. Adalsteinsson⁵ et al. collected data at varying echo times to sample the FIDs and used an FFT in the time dimension to sample the chemical shift spectrum. Sarkar⁶ et al. refined the reconstruction to reduce side lobes from truncations of the data in echo-time and spatial k-space domains.

We implemented a multi-echo, multi-interleaf spiral pulse sequence in the IDEA pulse programming environment on our Siemens 3T Tim Trio MRI scanner (Siemens Medical Solutions). Two methods for measuring T₂* will be discussed. A slow but reliable way of measuring T₂* is to make repeated measurement of the FID but instead of identical measurements, the echo-time is stepped by a fixed step size, typically 500us. After Fourier transformation, this results in a spectral window of 1000Hz. Our pulse sequence allows us to collect these data with single shot or multi-interleaf spiral k-space trajectories. Hardware limits, fields-of-view, resolution and maximum data acquisition time, dictate the choice of parameters used to generate the trajectories at runtime. This type of spiral scan can be run with a single or multiple echoes but we typically collect these with only a single readout window. By making the step size much shorter than the readout duration, the blurring effects of off-resonance can be significantly reduced. It should be noted that intra-voxel dephasing due to the susceptibility gradients will still result in local signal loss.

The alternate method for measuring T₂* is to collect multiple echoes. The number of echoes and times between echoes can be adjusted by using multiple interleaves. Increasing the number of interleaves shortens each readout period and the echo spacing can be reduced. The disadvantage of doing this is we violate Nyquist sampling in the echo-time dimension but a field map and multi-frequency reconstruction can be used to reduce blurring from off-resonance. Again, intra-voxel dephasing cannot be corrected and results in signal loss. However, shorter spiral readout times do reduce this signal loss.

Fig. 1 shows the results of a 25 minute single shot spiral scan that was acquired by incrementing the echo time by 500usec for each of the 256 frames. The spiral was a 2D single shot spiral with a 192mm FOV, a resolution of 64 by 64 with a slice thickness of 3mm. The slew-rate was limited to 120 mT/m/msec to avoid peripheral nerve stimulation and resulted in a slew-rate limited spiral with a maximum gradient of only 25 mT/m. Fig. 2 shows one of the 256 frames. Figure 3 shows the data after Fourier transformation and the off-resonance blurring is inherently corrected.

Figures 4 and 5 show the fitted M₀ and T₂* from 1mm isotropic six-echo 32-interleaf spiral data acquired of the whole head.

Various trajectories are possible for rapidly collecting data in k-space. Spiral has long been a reliable and robust method. We have explored a few variations of spiral imaging to make quantitative measurements of the brain in vivo in the presence of susceptibility gradients. Slow and tedious measurements are possible and future work will reduce the number of acquired echo-time steps.

Multi-echo, multi-shot imaging is also possible and interleaving the spirals reduces the time required for each readout. Readout time should be kept short to reduce the dephasing from off resonance when the echo time is not stepped. M₀ and T₂* were fitted to the multi-echo data but are not immediately comparable to the complete spectral measurements because T₂* depends on the local susceptibility gradients across the voxel. Voxel size affects T2*.