Tm. Keaveny et al., THEORETICAL-ANALYSIS OF THE EXPERIMENTAL ARTIFACT IN TRABECULAR BONE COMPRESSIVE MODULUS, Journal of biomechanics, 26(4-5), 1993, pp. 599-607
A theoretical analysis was performed to characterize potential experim
ental artifacts in conventional compression testing of trabecular bone
, where strains are based on the relative displacements of the two loa
ding platens. We assumed that the total experimental artifact for modu
lus was the SLIM of a damage and friction artifact and derived equatio
ns to describe these artifacts. The two unknown constants in these equ
ations were found using a combination of data derived from linear fini
te element analyses and in vitro uniaxial compression tests. Subsequen
t finite element analyses allowed estimation of the artifacts for a wi
de range of specimens (cube, 1:4 - 3:1 aspect ratio cylinders). If fri
ction is completely eliminated at the specimen-platen interface, the Y
oung's modulus of a 5 mm sized (1:1 aspect ratio dimension) specimen w
hich has a damage artifact due to machining may be underestimated by a
t least 45% regardless of specimen geometry; otherwise, the platens mo
dulus may vary from less than 30 to over 175% of the Young's modulus,
depending upon the specimen geometry and Poisson's ratio of the bone.
Increasing the specimen size reduces the artifact only slightly. Since
Poisson's ratio can be large for trabecular bone and is rarely known
a priori, the precision of the conventional compression test will, the
refore, be poor unless friction is completely eliminated al the specim
en platen interface. However, without friction at the interface. the p
latens modulus will always underestimate Young's modulus, thereby redu
cing the accuracy of this test. There was also evidence that the stren
gth may be affected by these artifacts. Taken together, these data sug
gest that the conventional compression test can be precise but is rare
ly accurate. and that inter-study comparisons should be made with caut
ion. Use of other methods such as ultrasound or direct attachment of e
xtensometers to material away from the platens may overcome problems w
ith accuracy. Finally, a protocol that eliminates friction and uses a
2:1 aspect ratio specimen may optimize precision.