This Is AuburnElectronic Theses and Dissertations

Noninvasive hepatic lipid quantification with magnetic resonance imaging and spectroscopy




Khan, Salaar

Type of Degree

Master's Thesis


Electrical and Computer Engineering


Excessive accumulation of intra-hepatocellular lipid (IHCL) can lead to one of the most common forms of chronic liver diseases in adults, the non-alcoholic fatty liver disease (NAFLD), that comprises a range of liver disorders including hepatic steatosis, an advanced stage of which can result in liver cirrhosis. Many diseases are known to be associated with the retention of IHCL including, but not limited to, obesity and type II diabetes. It is, thus, essential to quantify IHCL for early diagnosis and monitoring for an effective treatment. This study presents a non-invasive, robust and reproducible approach to quantify IHCL content using the single-voxel 1H magnetic resonance spectroscopy (MRS) and magnetic resonance imaging (MRI) at 3 Tesla (3T). The study is divided in two parts: the MRS and the MRI investigations. Both parts used the constructed peanut oil phantoms of known fat fractions (8%, 14%, 18%, 25%, 30%, 40%, 45%, 55%) to test for validation and accuracy. The results with the peanut oil phantoms were in close agreement with the known fat fraction values (<10% error on the fat fraction value). The in-vivo MRS study included fifteen volunteers whose IHCL data was obtained at three different voxel locations in the liver with T2 corrections using five echo times (24, 30, 35, 40, 50 ms). The study confirmed the obese volunteers with fatty liver, registering IHCL content beyond the 6% clinically considered normal range. The in-vivo MRI study included liver data with similar protocol from multiple sites: Auburn University, University of Mississippi, Pennington University and University of Alabama at Birmingham. Existing schemes to model fat and water for separation into images were used to compute fat fraction images on which region of interest (ROI) was marked for analysis.