Calibration of a semi-automated segmenting method for quantification of adipose tissue compartments from magnetic resonance images of mice

ObjectiveTo use an automated water-suppressed magnetic resonance imaging (MRI) method to objectively assess adipose tissue (AT) volumes in whole body and specific regional body components (subcutaneous, thoracic and peritoneal) of obese and lean mice.Materials/MethodsWater-suppressed MR images were obtained on a 7 T, horizontal-bore MRI system in whole bodies (excluding head) of 26 week old male C57BL6J mice fed a control (10% kcal fat) or high-fat diet (60% kcal fat) for 20 weeks. Manual (outlined regions) versus automated (Gaussian fitting applied to threshold-weighted images) segmentation procedures were compared for whole body AT and regional AT volumes (i.e., subcutaneous, thoracic, and peritoneal). The AT automated segmentation method was compared to dual-energy X-ray (DXA) analysis.ResultsThe average AT volumes for whole body and individual compartments correlated well between the manual outlining and the automated methods (R2 > 0.77, p < 0.05). Subcutaneous, peritoneal, and total body AT volumes were increased 2–3 fold and thoracic AT volume increased more than 5-fold in diet-induced obese mice versus controls (p < 0.05). MRI and DXA-based method comparisons were highly correlative (R2 = 0.94, p < 0.0001).ConclusionsAutomated AT segmentation of water-suppressed MRI data using a global Gaussian filtering algorithm resulted in a fairly accurate assessment of total and regional AT volumes in a pre-clinical mouse model of obesity.