Meeting Banner
Abstract #0195

Non-Parametric Quantification of Cerebral Haemodynamics from Dynamic Susceptibility Contrast MRI

SUMMA25Amit Mehndiratta1, Bradley J. MacIntosh2, David E. Crane2, Stephen J. Payne3, Michael A. Chappell3

1Institute of Biomedical Engineering , University of Oxford, Oxford, Oxfordshire, United Kingdom; 2Medical Biophysics, University of Toronto, Toronto, ON, Canada; 3Institute of Biomedical Engineering, University of Oxford, Oxford, Oxfordshire, United Kingdom

DSC-MRI analysis is an ill-posed inverse problem involving deconvolution of the observed MR signal with an AIF. The current standard SVD methods underestimate perfusion with an oscillatory residue function whereas the alternative vascular model-based (VM) approach permits only predefined shapes that may not be appropriate in pathology. Here we propose a deconvolution method that can estimate perfusion along with a physiological plausible residue function without any bias to a specific class of functional shapes. Our results showed the perfusion estimates were comparable to VM and the method formulation ensures physiologically realistic smooth functions.

Abstract PDF Presentation Video

Keywords

absolute accurate acquisition actual amplitude appropriate arrival arterial assumed assumptions avoiding axial belong best biophysics blood bolus brain cerebral class clinical complete constraints control crane critical cubic curves deconvolution degrees delay desirable develop disease draw estimating estimation exponential family fitting flow flows form formed freedom function functions generate generated goal good highly in vivo informative input introduce inverse journal kingdom known like linear maps matrix maxim medical medicine metabolism model models monotonically noise noon novel offers optimization oscillations oxford parametric pathological patient permits physiological plausible practical processing produces proposed realistic requiring residue response revealed sample scaled sensitive shape shapes simulated slice smooth solution solutions spline strict theory tissue tracer trio typical typically unction underestimate underestimated underlying validation variant vary whilst wide wing