Detection of dispersion from bolus-tracking MRI improves cerebral blood flow quantification: a simulation study
Zanderigo F, Cobelli C, Bertoldo A, Pillonetto G
University of Padova
Currently, CBF is estimated from the maximum of the dispersed deconvolved curve R(t) instead from the true R(t) at time t=0. This leads to underestimation of CBF. Thus, a deconvolution method able to detect the non dispersed component R(t) is needed. A nonlinear stochastic regularization method (NSR) has been proposed to account for smoothness and non-negativity of R(t). Here, the NSR ability in discriminating between non dispersed and dispersed components of the effective residue function is shown on simulated data. NSR shows to detect the degree of bias affecting CBF estimates thus improving CBF quantification.