E2351. MR Elastography Quantitative Liver Stiffness Comparison: Weighted vs Unweighted Mean
  1. Anshuman Panda; Mayo Clinic
  2. Justin Yu; Mayo Clinic
  3. Annelise Silva; Wright State University
  4. Matthew Doan; Mayo Clinic
  5. Alvin Silva; Mayo Clinic
MR Elastography (MRE) has been widely accepted as a noninvasive quantitative biomarker for liver stiffness. The process of generating liver stiffness measurements remains a manual process requiring expert oversight in drawing accurate ROIs. Additionally, it has been suggested that the sizes of the ROIs obtained for MRE can differ significantly so the weighted mean stiffness may be more accurate than the unweighted mean stiffness. Calculating the weighted mean stiffness requires an additional step of recording each ROI area, and the potential for error increases because the weighted mean formula doubles the independent variables as the unweighted formula. In contrast, the unweighted mean stiffness can be calculated simply by drawing the ROIs and averaging the ROI stiffnesses, which is faster and less prone to calculation error. To our knowledge, this comparison has not been previously performed, and thus the objective of this study was to evaluate whether the weighted or unweighted mean result in differences in clinical reporting of liver stiffness values as measured with MRE.

Materials and Methods:
Liver stiffness measurements (LSM) were calculated on MRE cases performed over last six months at our institution. These cases included both 1.5T and 3T field strengths. Only cases where high confidence region was present to draw an ROI on all four slices through the liver were included in the analysis. All ROIs were drawn by the interpreting radiologist. A mean LSM in kPa (m) and ROI size in mm2 or cm2 (w) is generated for each slice. The overall liver stiffness for the exam is then calculated as the mean value of the measurements. The unweighted mean is the simple arithmetic mean of the LSM values from the ROIs obtained. The weighted arithmetic mean (AMw), which accounts for the size of each ROI, was calculated with the following formula: AMw = (m1w1 + m2w2 + m3w3 + m4w4) ÷ (w1 + w2 + w3 + w4). *mean LSM in kPa (m) and ROI size in mm2 or cm2 (w).

In total, 161 cases met the inclusion criteria. In the analyzed cases, the unweighted mean liver stiffness ranged from 0.99 KPa to 11.48 kPa, and the weighted mean liver stiffness ranged from 1.02 kPa to 11.49 KPa. A two-tail paired t-test of the weighted versus unweighted mean values yielded a p-value of 0.567.

It has been generally recommended that the weighted mean corrects for the different sizes of the ROIs on each slice which may offer better performance for liver stiffness measurements obtained with MR elastography, although no published data have demonstrated the superiority of either approach. Our analysis indicates that there is no strong evidence for any significant difference between the weighted and unweighted mean liver stiffness. Using the unweighted mean can allow for faster calculation and reporting of liver stiffness measurement without statistically significant decline in performance or accuracy.