Meeting Banner
Abstract #4036

MR-Liver Fat Volume Fraction Quantification Using a Magnitude-Based Technique with Independent Fat and Water T2* Estimations, T1-Related Bias Correction and Accounting for Fat Multiple Resonances

Benjamin Leporq1, Helene Ratiney1, Herve Saint-Jalmes2, 3, Frank Pilleul1, 4, Olivier Beuf1

1CREATIS; CNRS UMR 5220; INSERM U1044; INSA-Lyon; UCBL Lyon 1, Villeurbanne, Rhne-Alpes, France; 2LTSI; Inserm U642; Universit Rennes 1, Rennes, Bretagne; 3CRLCC; Centre Eugne Marquis, Rennes, Bretagne, France; 4Hospices civils de Lyon; CHU Edouard Herriot, Lyon, Rhne-Alpes, France

In Western countries, Non-Alcoholic Fatty Liver Disease (NAFLD) is the most common cause of chronic liver diseases. Because NAFLD can evolve into Non-Alcoholic Steato-Hepatitis and may lead to liver fibrosis up to cirrhosis, a clinical follow-up of NAFLD would be valuable. This work presents a MR magnitude-based technique for liver fat volume fraction quantification with independent fat and water T2* estimations, T1-related bias and spectral complexity of fat corrections using a two-angles multiple gradient-echo acquisition. The method described was applied in-vivo at 1.5T in a prospective study and at 3.0T on healthy subjects. It was additionally compared with 1H MRS on fat-water phantoms.

Abstract PDF E-Poster Video

Keywords

accounting acquisition acquisitions affected agar agreement aims alcoholic allowed almost amplitude apparent assessment assuming axial bandwidth bias biopsy breath cause cells chronic clinical coefficient coil combined common complications components computed computing confirmed consisted containing corrected correcting correction correlated density described disease diseases domain done doped echoes eddy emulsions encountered enrolled equally errors especially estimations exponential fatty features fits flash fraction frank function furthers good grade gradient greater healthy hence histological histology ideal identical important in vivo independent individual inherent interference intracellular invasive involve least limitation linear lipid liver located magnetizations magnitude matrix medical mild minor mixture model morbidity moreover need nevertheless nonlinear parametric peak phantom pixel platforms post proceedings processing prospective proton pure quantification quantified radiology readout recombined reeder regression related repeated respectively risk robustness sampling scanned scored separation simplified since slice spectra spectrum squares stages start steam stratified strongly subjects suitable systematically torso twice validations versatility volume volunteers water whose women years