A deep-learning approach to accelerated T1-T2-relaxationcorrelation imaging

摘要

Introduction: Multidimensional correlation imaging is a powerful technique to quantify microstructural tissue compartments.[1] However, a robust solution of this ill-conditioned problem comes at the cost of long acquisition times and the need for advanced optimization algorithms, hindering its clinical application.[2,3] For the first time, we jointly address these inherent limitations and present a neural network- based estimation of voxelwise T1-T2-spectra that accelerates correlation imaging in two ways.