Fully Automated Quantification of Subcutaneous & Visceral Abdominal Adipose Tissue using Water & Fat Acquisition & Graph Cuts

Vitali Zagorodnov¹, Sarayu Parimal², Michael W. L. Chee²

We developed a fully automated algorithm for quantification of visceral (VAT) and subcutaneous (SAT) adipose tissue. Our algorithm is based on water and fat acquisition (Dixon sequences), and improves on previous work by utilizing fat-fraction images (for fat/non-fat separation) and graph cuts (for separation of VAT and SAT compartments). In our evaluation, the algorithm achieved excellent correlations with manual measurements, and outstanding reproducibility for both SAT and VAT (coefficient of variation 1.5% and 2.3% respectively).