WE‐C‐J‐6B‐01: Fast 4D Imaging: Breaking the Speed Limits in MR and CT

Abstract

Advances in MRI and CT have made it possible to produce time‐resolved series of 3D volume images. In CT the increase in detector rows to the present value of 64 ‐ or even 256 ‐ has brought a significant increase in image quality and reduced acquisition times. The emergence of volumetric cone beam systems employing flat panel detectors also offers the possibility of rapid 4D acquisition. In MRI, 4D sequences have been facilitated by the introduction of multi‐coil parallel imaging methods and several acquisition schemes using novel k‐space sampling schemes. The vast majority of MRI applications use Cartesian acquisitions employing phase encoding. This line‐by‐line scan through k‐pace produces spatial resolution that is linearly related to acquisition time. This is basically a k‐space speed limit imposed by the Nyquist theorem. For dynamic applications this tradeoff between temporal and spatial resolution is unacceptable. When highly undersampled radial acquisitions are used, spatial resolution is determined by readout resolution rather than the number of k‐space lines acquired. The price paid for violating the Nyquist theorem is streak artifacts. However, when the data acquisition is done using a time‐resolved 3D radial acquisition called VIPR(Vastly undersampled Isotropic PRojection imaging), the streaks are often tolerable and have permitted acceleration factors up to 60 relative to Cartesian acquisition for applications such as phase contrast imaging where background signals are cancelled and streaks can only be generated by the vessels of interest. Undersampled cone beam acquisition using an array of x‐ray sources combined with large area detectors has been proposed as a means of increasing the temporal resolution of CT and appears to have similar properties to 3D undersampled MRI acquisition.