X-f choice reconstruction of dynamic undersampled MR data applied to DCE-MRA
Malik S, Hajnal J, O'Regan D, Schmitz S
Hammersmith Hospital, Imperial College London
Dynamic undersampling in the k-t domain causes aliasing in the conjugate x-f domain, which must be resolved. Based on the observation that signal distribution in x-f space is sparse, x-f choice reconstruction proceeds by reducing the size of problem to solve for each aliased voxel group to locally reflect the true amount of signal overlap. Overlaps are resolved using coil sensitivity data. The problem size is inferred from a model x-f space constructed directly from undersampled data using a k-means clustering approach, meaning no training scan is required. Presented are images from in vivo application and a validation study.