Sensitivity of magnetic-resonance current-density imaging

This paper analyzes the sensitivity of magnetic-resonance current-density imaging (CDI) to random noise and systematic errors with the goal of providing a protocol for achieving a targeted noise performance. All analyses are checked with an electrolytic phantom designed to create a uniform current density or with an equivalent computer model. CDI's sensitivity properties are split between low and high current-time regimes where random noise or systematic error, respectively, dominate. Large current-time products introduce nonlinear signal loss in the magnitude image. Image alignment errors, spatial linearity, pixel size calibration, and current-dependent ringing can exceed random noise. In the low current-time limit, random noise dominates and is proportional to the inverse cube of the planar pixel dimensions. A spin-echo sequence minimizes the current-density noise for current integration times and echo times in the neighborhood of T 2. The relation between current-density noise and the MR image resolution and signal-to-noise ratio is formulated, allowing one to predict the noise for experimental planning. The lower practical limit of sensitivity is estimated to be 0.1 A/m 2 for a 1.5 × 1.5 × 5 mm voxel.