Figure 1.

Data acquisition overview. (a) Pulse sequence diagram. (b) Cartesian sampling masks generated using the variable-density and radial view-ordering (VDRad) design. (c–f) Single dataset acquired in the multi-dimensional space reconstructed with different methods of data binning where each data block consists of three-dimensional spatial (x, y, z) and flow encoding (f). In (a) different dimensions are illustrated: dynamic contrast enhancement (t), respiratory phase (r), and cardiac phase (c). For volumetric flow quantification, four different flow-encoding gradients are used to sensitize the acquisition to four different velocities. The flow-encoding gradients provide the setup to intrinsically measure motion using Butterfly navigators. In (b) VDRad determines the order in which each (k _y, k _z)-sample is collected using the golden-ratio angle increment (~137.5°). The same samples can be divided into different number of sampling masks where fewer masks correspond to lower subsampling reduction factors (from R of 6.8 to 2.1). This property is important for retrospective binning of the data. The variable-density sampling provides ideal source data for compressed sensing reconstructions. Conventional 4D flow depicted in (c) ignores the r, c, and t dimensions. XD flow extends this same single dataset into higher-dimensional-space for highlighting different clinical indications. Three different examples of this flexibility is illustrated in (d–f).