Identification of an optimal threshold for detecting human brown adipose tissue using receiver operating characteristic analysis of IDEAL MRI fat fraction maps

Publication date: September 2018
Source:Magnetic Resonance Imaging, Volume 51
Author(s): Terence A. Jones, Sarah C. Wayte, Narendra L. Reddy, Oludolapo Adesanya, George K. Dimitriadis, Thomas M. Barber, Charles E. Hutchinson
PurposeLower fat fraction (FF) in brown adipose tissue (BAT) than white adipose tissue (WAT) has been exploited using Dixon-based Magnetic Resonance Imaging (MRI) to differentiate these tissues in rodents, human infants and adults. We aimed to determine whether an optimal FF threshold could be determined to differentiate between BAT and WAT in adult humans in vivo.MethodsSixteen volunteers were recruited (9 females, 7 males; 44.2 ± 19.2 years) based on BAT uptake on ¹⁸F-FDG PET/CT. Axial 3-echo TSE IDEAL sequences were acquired (TR(ms)/TE(ms)/matrix/NEX/FoV(cm) = 440/10.7–11.1/512 × 512/3/30–40), of the neck/upper thorax on a 3T HDxt MRI scanner (GE Medical Systems, Milwaukee, USA), and FF maps generated from the resulting water- and fat-only images. BAT depots were delineated on PET/CT based on standardized uptake values (SUV) >2.5 g/ml, and transposed onto FF maps. WAT depots were defined manually within subcutaneous fat.Receiver operating characteristic (ROC) analyses were performed, and optimal thresholds for differentiating BAT and WAT determined for each subject using Youden's J statistic.ResultsThere was large variation in optimal FF thresholds to differentiate BAT and WAT between subjects (0.68–0.85), with great variation in sensitivity (0.26–0.84) and specificity (0.62–0.99). FF was excellent or good at separating BAT and WAT in four cases (area under the curve [AUC] 0.84–0.92), but poor in 10 (AUC 0.25–0.68).ConclusionAlthough this technique was effective at differentiating BAT and WAT in some cases, no universal cut-off could be identified to reliably differentiate BAT and WAT in vivo in adult humans on the basis of FF.



https://ift.tt/2IGHeZP