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
Results
The mean FF value was 14.4% (SD = 9.2 / Min = 4.1% / Max = 54.8%).
There was a good correlation between the degree of steatosis and the FF by the PDFF method,
estimated in 0.72.
The cut-off value considered ideal for steatosis was 5.4% with AROC of .98 for the distinction between normal and fat liver patients (Fig 4).
The mean value of R2* was 67.5 s-1 (SD = 42 / Min = 34.5 s-1 / Max = 384 s-1).
The ideal cutoff value for detection of siderosis (in correlation with histology) was 68.8 s-1(AUROC = .79).
The correlation between FF obtained by the PDFF method and FD by conventional Dixon was very high,
estimated at 0.96,
with p <0.01.
A correlation between R2*,
FF and FD was also identified,
but with a low intensity,
being estimated at 0.51 and 0.46,
respectively.
Regarding the influence of the R2* values in the analysis,
there was an excellent correlation between the fat quantification methods in the first 3 quartiles (R2* <69.9 s-1),
higher than 0.95 with p <0.001.
Only for R2 * values ≥ 69.9 s-1,
the correlation was lower than the others,
but still significant,
with an estimated value of 0.87 and p <0.001 (Table 1).
These findings demonstrate a good correlation between conventional steatosis quantification methods using Dixon gradient echo sequences and advanced proton fat density methods,
allowing the use of the former,
if the latter are not available,
as long as there are no signs of iron deposition in the hepatic parenchyma.
This same quantification could be used in the control of these patients,
even if more advanced techniques were acquired later.
It should be noted some important limitations in the study:
- The study involved only NASH patients,
which could limit its use in other patients,
but this impact should be minimal,
since patients' steatosis and R2* values were very broad.
- Only one manufacturer's machine was used,
which could impact the use of other Dixon sequences.
This is related to variations among companies in the acquisition of three-dimensional Dixon sequences and could impact the correlation between more advanced techniques.
- A 3T machine was used,
which limits the results for this type of equipment,
based on different R2* values in 1.5T MRI’s.
Generally,
the values of R2* in 1.5T equipment are about a half of those obtained in 3T,
which could imply a limit for the correlation in 1.5T of 35s-1,
but this aspect is only speculative at this point.
- The values of R2* used were obtained with the use of PDFF processed images,
being different from those obtained in conventional relaxometry techniques,
which could limit the threshold values used.
- 5Only one FF postprocessing technique was used,
but this aspect should have a small impact in the analysis,
since there are already papers showing a good correlation between distinct methods of steatosis quantification with different PDFF techniques