Do not use Structure Model Index (SMI) for measuring plates and rods

[Michael Doube]

Dear all,

SMI has been included in BoneJ since its early releases, because it was a standard measurement used in the field.

While implementing it, it became clear to me that the basic premise of SMI that the entire surface is convex could not possibly be true for trabecular bone.

So, from the very beginning, BoneJ's SMI implementation reported the two values that contribute to SMI: SMI+ and SMI-.

With help from Phil Salmon from Bruker microCT, and academic colleagues in Sweden and the USA, I showed that SMI usually doesn't work on real bone geometries. The concave parts taking up as much as 70% of the bone surface mean that the SMI value you get as an output for your bone sample is as good as meaningless for interpreting plate/rod geometry. For more details please see our paper in Frontiers in Endocrinology "Structure model index does not measure rods and plates in trabecular bone" http://dx.doi.org/10.3389/fendo.2015.00162

And - please don't use SMI in your science!

Michael