Model-based deconvolution for DSC-MRI: A comparison of accuracy, precision, and computational complexity of transit time distributions

Abstract

To identify the optimal model-based deconvolution process for DSC-MRI, four models of transit time distribution (TTD) were compared in terms of goodness and stability of fit, consistency of perfusion estimates, computational complexity, and robustness against noise. Although all models gave similar fits, the gamma function converged faster and more consistently to the global minimum, regardless of the initial guess. Moreover, it gave more accurate and precise perfusion estimates in the presence of noise. We conclude that the gamma function is the most suitable TTD model for perfusion analysis, and may prove useful in urgent clinical situations and multi-centre studies.