Ludovic De Rochefort1, Hongchen Wang2, Paulo Loureiro de Sousa3, Jean-Paul Armspach3
1Univ Paris-Sud, Orsay, France; 2Univ. Paris Sud - CNRS, UMR 8081, IR4M, Orsay, France; 3Universit de Strasbourg, CNRS, ICube (UMR 7357), FMTS, Strasbourg, France
a full processing of multi-echo gradient-echo data brain is described to efficiently extract the field inhomogeneities inside the brain while removing background effects. First, phase maps including background effect are unwrapped in time using a fast Fourier transform, and then fitted to an affine model to estimate the field. A T1-weigthed scan is then used to segment the brain. Internal field within the brain is finally efficiently extracted using harmonic filtering, i.e. by solving Laplace equation with adequate boundary conditions. Field wrapping in space is removed in the process using the modulo function. Internal field and QSM obtained in vivo at 3T are presented, and compared to R1 (=1/T1) and R2* = (1/T2*) maps.