Figure 7.
Sample size estimates with 80% power when mixing 1.5 T and 3 T scans. The lowest sample size number (n80), indicating greatest power, is found in the 1.5 T group when using the statistically-defined ROI based on 1.5 T scans. The worst performance arises when analyzing the 3 T scans using the 1.5 T-specific statistical ROI. Power can be increased in the 3 T group if a statistically-defined ROI based on 3 T scans is used instead, perhaps because some voxels with artifacts are avoided. Mixing 1.5 and 3 T scans does not affect the n80 adversely, and sample sizes for mixed scanner designs lie in between those using 1.5 or 3 T exclusively. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
