Pancreas ectopic fat: imaging-based quantification

The increasing incidence of obesity and nonalcoholic fatty liver disease worldwide strongly suggest that pancreatic fat will also become a common entity. Pancreatic ectopic fat can be seen histologically as intralobular and interlobular accumulation of adipocytes. Small amounts of pancreatic fat are benign and are a relatively common finding in histology and imaging. Greater amounts of pancreatic fat have been linked with insulin resistance, and ectopic fat accumulation and inflammation in the pancreas are among the early manifestations of chronic pancreas disease. For these reasons, pancreatic fat imaging has received increasing interest in the past few years. Magnetic Resonance (MR)-based methods are increasingly considered gold-standard for noninvasive pancreatic fat quantification. Useful imaging biomarkers obtained from MR include pancreatic volume, shape features, proton density fat fraction, and relaxometry metrics. Regional assessment of pancreas pathology is also important, for instance, by head, body, and tail. However, conventional sampling strategies are not likely to capture the heterogeneous fat deposition in the pancreas accurately. Similarly, commonly used acquisition protocols for liver imaging may need to be repurposed. The availability of large datasets with both imaging and nonimaging data will further facilitate the development of machine learning for pancreas segmentation, biomarker discovery, and shape characterization.