N onalcoholic fatty liver disease has a global prevalence of 25% and is defined as the intracellular accumulation of fat in the liver parenchyma exceeding 5% in the absence of competing causes such as chronic viral hepatitis, druginduced steatosis, or other chronic liver diseases such as autoimmune hepatitis, hemochromatosis, or alcohol abuse (1,2). Nonalcoholic fatty liver disease is considered an important cause of chronic liver disease, such as nonalcoholic steatohepatitis and cirrhosis, and even hepatocellular carcinoma (3). People with nonalcoholic fatty liver disease tend to also have obesity, diabetes, dyslipidemia, and hypertension, and are at higher risk of cardiovascular disease (1). Accurate and reliable quantification of liver fat content is critical for the diagnosis, treatment, and monitoring of nonalcoholic fatty liver disease.Liver biopsy has long been the reference standard for assessment of liver fat content, although its application is restricted because of its invasive nature and high cost (4). Noninvasive techniques include US, CT, and MRI. US is an inexpensive and convenient diagnostic tool for nonalcoholic fatty liver disease, but it is semiquantitative and relatively insensitive to individuals with liver fat content less than 20% (5-7). Traditionally, CT was considered accurate for moderate-to-severe steatosis but insensitive to mild steatosis (8,9), and the results are susceptible to variable scanning conditions such as different tube voltages and CT scanners from different manufacturers (9,10). MRI methods, including proton MR spectroscopy and the emerging technique of chemical shift-encoded (CSE) MRI, are regarded as the most accurate noninvasive techniques for the evaluation of liver fat (11)(12)(13)(14)(15). Proton MR spectroscopy and CSE MRI quantify liver fat directly in terms of the proton density fat fraction (PDFF), defined as the ratio of the signal strength from fat to the total signal from fat and water (11,16).