Search papers, labs, and topics across Lattice.
2
0
1
61
Prior MRI scans can slash reconstruction time for new scans by 4x without sacrificing quality, thanks to a novel attention mechanism that handles misregistration and phase shifts.
Self-supervised learning in MRI can be dramatically enhanced by tailoring pretraining objectives to the specific structures of downstream tasks, with MAE showing superior performance when focused on high-frequency details.