How to reduce inter-rater variability in the analysis of myosteatosis?

Hello! I’m working on a project evaluating muscle mass in CT images of breast cancer patients undergoing neoadjuvant chemotherapy. In addition to the cross-sectional muscle area, we’re also interested in analyzing the myosteatosis area within the studied muscles. To do this, we’re using the Segment Editor and setting the fat range between -190 and -30 HU, with quantification done via the SandBox - Cross Sectional Area extension.

However, when performing the analysis with two independent raters to calculate the coefficient of variation, we noticed that the standard deviations between the evaluations were quite high. Does anyone have any suggestions on how to reduce this variability and improve the consistency of the results? Any insights on this topic would be greatly appreciated!

All the best,
Marcelo