Abstract #1779

Model-Free Maximum Likelihood Deconvolution (MLD): A Novel Perfusion Analysis Method in MRI to Calculate Myocardial Blood Flow

Omar El-Sherif¹, Robert Stodilka^1,2, Benoit Lewden¹, Michael Jerosch-Herold³, Terry Thompson^1,2, Frank Prato^1,2

¹University of Western Ontario - Lawson Health Research Institute, London, Ontario, Canada; ²Diagnostic Radiology, St. Joseph's Health Care, London, Ontario, Canada; ³Radiology, Brigham & Women's Hospital, Boston, MA, USA

We introduce a novel contrast enhanced perfusion analysis method maximum likelihood deconvolution (MLD) to estimate myocardial blood flow. MLD is designed to overcome difficulties that arise due to noise without compromising accuracy. MLD is evaluated using computer simulations and clinically in three volunteers. Simulation and clinical tests demonstrated that MLD is numerically stable and accurately estimates perfusion at clinical contrast:noise levels for cardiac perfusion values. MLD models dispersion of the contrast agent between the arterial and tissue curves, which is not usually accounted for by standard deconvolution analysis.