Analysis of systematic and random error in MR volumetric flow measurements

The spatial aspects of error in 2D MR cine phase-velocity mapping are considered in order to define acquisition strategies which will minimize error in measuring volumetric flow. Error was separated into two categories: systematic and random. Potential sources of systematic error examined were intravoxel phase dispersion (IVPD), partial volume effects, misalignment of flow axis and flow-encoding gradients, and improper choice of vessel voxels for flux calculations. Random error was addressed using analysis of propagation of variance. Analytical expressions for sources of error were derived; and computer models were used to test the analytical models. Flow phantom studies examining error in MR volumetric flow measurements were performed and compared with error predicted by the analytical models. Expected error in several clinical situations of interest was then derived to find appropriate acquisition strategies. Spatial resolution, signal to noise ratio, velocity sensitivity and the ratio of the modulus of moving isochromats to that of static isochromats were found to be the most important parameters in controlling error and were found to cause competing effects with respect to systematic and random error.