Figure 3. Slice Thickness Affects Frequency Maps.

With linear gradients, the slice thickness would not be expected to affect the frequency map generated by a standard field mapping sequence. In our uniform phantom (a), the slice at isocenter yields the same frequency map (b) for the z²-½(x²–y²) gradient regardless of slice thickness. There is no bias introduced as the slice is widened to 4, 6, 8, or 10mm (c–f). However, at z=2.5cm (g), the analytical expression predicts that phase evolution from intravoxel dephasing becomes significant. If this evolution is interpreted as evolution of the bulk, the observed frequency of the slice becomes a function of the slice thickness (h–l). A linear fit to the exact expression in Eq. [3] predicts biases of −100, −400, −550, −900Hz for these thicknesses, in good agreement with the data.