Authors:
H. Momo Jeufack, L. Vidarsson, C. Piron, D. B Plewes; Toronto, ON/CA
DOI:
10.1594/ecr2010/C-3263
Results
Within an image, the g-factor is dependent on the specific coil array, the spatial location and the acceleration factor. The SNR loss is commonly the main limiting factor in acceleration speed due to rapid growth of the g-factor beyond an acceleration factor of 4 in one direction. However, accelerations in two directions have a multiplicative total acceleration factor that cannot be achieved along one direction without increasing the geometry factor and, hence, the local noise amplification. This would be beneficial on 3D imaging techniques that have two phase-encoding directions. Figure 3 shows the noise amplification when acceleration is done in one direction (LR) and in two directions (LR/SI).
Fig.: Effect of the acceleration direction on g-factor. A) Acceleration applied in one direction (LR=8). The g-factor is amplified in the chest wall region and extends to the breast area. B) Acceleration in two directions (LR = 4 and SI =2). The multiplicative total acceleration factor causes a decrease of the g-factor contributing to less local noise amplification.
This result shows a considerable increase in noise in the chest wall and part of the breast when R=8 is applied in one direction (LR), while for the same acceleration distributed along two directions (LR/SI), the noise amplification is reduced. Based on these results, one-dimensional acceleration factors higher than 4 were not used.
Using an acceleration factor of 4 resulted in a mean g-factor of 1.25 well within acceptable limits. Acceleration factor of 8 (LR = 4 and SI =2) resulted in a mean g-factor of 1.75. We used R=3 as the baseline, as this would represent the maximum acceleration used in a clinical setting. Figure 4 shows g-factor maps and corresponding volunteer images. High resolution images did not show any parallel imaging artifacts, For R=8, that is 4 in the LR and 2 in the SI, the scan time was 20s.
Fig.: The first column represents the acceleration factor. The second column corresponds to the actual high resolution, fat suppressed breast images. The g-factor color scale is provided for reference. The g-factor map indicates that there is no significant change from R=4 to R=4x2. The average g-factor over the volume was 1.25 vs. 1.75. The volunteer images do not show any PI artifacts.
Whereas parallel imaging involves a penalty in SNR, high resolution images did not show any parallel imaging artifacts using the high-sensitivity 16 channels breast coils at 3T.