Evaluating Childhood Obesity: Magnetic Resonance-Based Quantification of Abdominal Adipose Tissue and Liver Fat in Children

Abstract

The purpose of this study is to establish and validate a magnetic resonance (MR)-based fat quantification package that provides an accurate assessment of abdominal adipose tissue and liver fat in children. Ex vivo trials with a torso model and water-oil mixtures are conducted. Abdominal adipose tissue (AAT) is covered by magnetic resonance imaging (MRI) using a fat-selective sequence and is analyzed by a plug-in based on the open source software ImageJ. Liver fat (LF) is measured with localized 1H Magnetic Resonance Spectroscopy (1H MRS) and the jMRUI (java-based Magnetic Resonance User Interface) software package. Evaluation of the clinical methodology involved a study of 10 children in this feasibility study (mean age and body mass index: 13.3 yr; 33.3 kg/m²). To evaluate the method's validity, reference measurements were performed. Ex vivo trials with the torso model showed that adipose tissue was measured appropriately with a systematic underestimation by 9.3 ± 0.2 % (0.32 ± 0.064 kg). Coefficients of variation for both intra- and inter-observer measurements ranged between 0 - 2.7 % and repeated analyses showed significant equivalent results (p < 0.01). The lipid content obtained by 1H MRS ex vivo revealed significant equivalence with the predefined fat content in water-oil mixtures (p < 0.01). In vivo, the homemade plug-in significantly overestimated the AAT, with the visceral adipose tissue being most affected (+ 15.7 ± 8.4 %). Although an overestimation of the AAT by the presented plug-in should be taken into consideration, this children-friendly package enables the quantification of both LF and AAT within 30 min on a freeware-based platform.