Keywords:
Liver, Gastrointestinal tract, Abdomen, MR, Diagnostic procedure, Imaging sequences
Authors:
H. Leao Filho1, D. Belmonte1, T. Miranda1, A. Bordini1, L. Fabiana1, P. Clark2, W. Chua-anusorn2, M. de Souza Rocha1; 1Sao Paulo/BR, 2Perth/AU
DOI:
10.26044/ecr2019/C-1564
Methods and materials
A total of 102 patients with NASH diagnosed by percutaneous biopsy were analyzed.
Patients underwent a dedicated MRI study on a 3T device (Phillips Achieva) with a surface coil,
less than 6 months after the biopsy.
14 volunteers with no history of liver disease and normal laboratory tests for hepatic function were also included.
The assessment of iron and fat deposition in the liver was made through a multi- spoiled gradient echo sequence (MSGE),
where magnitude data were acquired.
This sequence lasts around 13 seconds and is obtained during an apnea,
involving the entire extension of the liver.
After the acquisition,
the images were processed by a specific software (Dive-In® - Magnepath - Perth,
Australia),
which calculates the percentage of triglycerides in the selected liver area,
correcting for several confounding factors such as R2 *,
T1 and the existing multiple fat peaks.
The result is a percentage fat map of the liver,
which can range from 0 to 100% (Fig 1).
A conventional volumetric Dixon sequence was also obtained during an apnea which resulted in IN-PHASE (IP),
OUT-PHASE (OP),
Water and Fat images,
with similar parameters to those obtained in the above-mentioned MSGE sequence.
For the analysis of fat and iron,
regions of interest (ROI's) involving the largest area of the liver in a sectional image were drawn,
excluding major vessels,
resulting in values of the fat fraction (FF,
in%) and R2 * (in s-1).
In the IP and OP images,
the same ROI's were drawn (Fig 2) and fat deposition (FD) was calculated using the following equation (Fig 3):
We use Kolmogorov-Smirnov tests to evaluate the normality of the variables and the Spearman correlation coefficient to analyze the correlation between FF and FD.
In order to assess the influence of R2 *,
FF and FD,
correlation tests were performed for four distinct strata of R2 *,
based on their quartiles: <47.6,
47.6 | - 58.4,
58.4 | - 69.9,
=> 69.9 .
All tests took into consideration a bidirectional α of 0.05 and 95% confidence interval (CI) and were performed with IBM SPSS 25 software (Statistical Package for the Social Sciences) and Excel 2010 (Microsoft Office).