Stratec images typically have pretty low resolution, and can’t reliably be used to segment trabeculae into the binary images that Thickness needs because the pixel size is close to or larger than trabecular thickness (Tb.Th). It’s giving you a result but the input image is probably inappropriate to measure Tb.Th and Tb.Sp. You might have enough resolution to measure cortical thickness, however, provided that you generate binarised images containing only cortex in the foreground. In this case you can measure “trabecular thickness” but your “trabecula” represented by Tb.Th numbers is actually the cortex.
Presumably negative results for SubCutFat density are incorrect, because that would require one of mass or volume to be negative. @tjrantal may be able to comment further.