Meeting Banner
Abstract #3907

Adaptive Neural Network for Direct Quantification of Longitudinal Relaxation Rate Change (δR1) in T One by Multiple Read Out (TOMROP) Sequence

Hassan Bagher-Ebadian1,2, Meser M. Ali3, Ali Seyd Arbab3, Malek Makki4, Siamak P. Nejad-Davarani1,5, Sawyam Panda1, Quan Jiang1,2, James R. Ewing1,2

1Neurology, Henry Ford Hospital, Detroit, MI, United States; 2Physics, Oakland University, Rochester, MI, United States; 3Radiology, Henry Ford Hospital, Detroit, MI, United States; 4Diagnostic Imaging, University of Children Hospital of Zurich, Zurich, Switzerland; 5Biomedical Engineering, University of Michigan, Ann Arbor, MI, United States

Direct and accurate measurement of the temporal change in the longitudinal relaxation rate after injection of a paramagnetic contrast agent has become increasingly important in MR perfusion studies. The Look-Locker (LL) sequence provides accurate T1 estimates which generally accomplished by nonlinear multi-dimensional curve fitting. However, these fitting methods are sensitive to initial guesses, and the errors of the parametric estimates. Herein, two Adaptive Neural Networks (ANNs) were trained using an analytical model of the LL signal. The model-trained ANNs were applied to the MR data acquired from animal model and human. Results were also compared with those of conventional methods.