Mbirn: MBIRN Protocols

Introduction

This page details protocols that are "morphometry BIRN approved", that is, they have been optimized either for raw signal-to-noise ratio or for reproducibility and accuracy of parameters that are calculated from the raw data (for example, the apparent diffusion coefficient calculated from diffusion-weighted imaging data).

We list both sequence parameters that have been optimized for a specific platform as well as parameters that could be used across platforms (for example, a multi-site trial that includes scanners from different manufacturers with different versions of software.

• Brain Morphology Structural MR Scan Protocol

Sequence: 3D spoiled gradient echo (FLASH type)
Slice orientation: Sagittal
No partial Fourier
NEX=1
FOV= 25cm
Phase matrix=192
Readout matrix=256
Number of Slices: 124 (no skip, whole head coverage, avoid wrap around)
Slice Thickness: 1.3mm-1.5mm
TE= minimum allowed with centered (symmetric) echo
TR=20 ms
Slab Selection: No (i.e., hard RF pulses)
scan 1: flip angle 30 degrees
scan 2: flip angle 5 degrees
scan 3: flip angle 20 degrees (desired for calibration, optional for routine scans)
scan 4: flip angle 3 degrees (desired for calibration, optional for routine scans)
Acquisition time per scan: about 8min 12sec

• Diffusion Tensor Imaging (DTI) Protocol

six diffusion gradient directions + one b=0 image
b = 0, 1000 s/mm2 (this is roughly the optimal b-value for brain imaging)
TR = 6 sec
TE = min for full acquisition
Number of slices = minimum needed to cover entire brain
Orientation = axial
Slice thickness = 3.0 mm
Slice gap = 0 mm
Want 3x3x3 isotropic voxels so either: FOV = 192 mm and 64x64 or FOV = 384 and 128x128.
4 data sets per subject
No zero filling or interpolation
Acquisition time per scan: 5-7 min

Things to consider in the aquisition of DTI data:

• It is not advisable to average scans together before the tensor calculation due to the presence of motion and physiological noise.
• There is a tradeoff between the number of diffusion gradient directions and the number of full data sets acquired in a given time.
• Multiple full tensor data sets should be acquired so images corrupted by motion can be replaced.
• It is better from a noise-insensitivity standpoint to acquire multiple acquisitions of two different b-values rather than single acquisitions of multiple b-values. (call these b=0 and b=large in what follows)
• The signal-to-noise ratio for the b=large image will be less than that for the b=0 image. Therefore, multiple copies of the b=large image should be acquired. Each of these can be treated as separate measurements in the calculation of the tensor.
• For isotropic samples the optimal ratio of the number of images acquired at each b-value is 1:3.6 (b=0:b=large)
• For anisotropic samples (such as white matter) the optimal ratio increases to 1:5.6 which can be rounded up to 1:6. Therefore the number of acquisitions at the b=large should be 6 times that of the number of acquisitions at b=0. (Jones 1999)
• The SNR that one needs for a given DTI experiment depends upon what one is trying measure. If one is trying to characterize regions of high anisotropy then one can live with somewhat lower SNR that if one is interested in regions of low anisotropy.
• The distribution of gradient directions in space is related to the noise performace of the tensor calculation. The electrostatic repulsion model for determining the distribution of gradient directions (Jones, 1999) gives good noise performance as does the Downhill Simplex Method given in (Skare, 2000). Gradient direction sets calculated from the electrostatic model are often referred to as "Jones#" where # is the number of gradient directions (for example, Jones30 for the 30 direction set.)

references

• Jones, DK, et al., MRM 42:515-525 (1999).
• Skare S, et al., JMR 147:340-352 (2000).

Other Protocols to Consider

The Alzheimer's Disease Neuroimaging Initiative (ADNI) has also published protocols for morphometric imaging. A summary document can be found here. ADNI has also published individual protocols for many different scanners from most of the major vendors. This list can be found here.