Oral Presentation ANZBMS-MEPSA-ANZORS 2022

Contribution of local density variation to micro-finite element analysis in pediatric HR-pQCT (#92)

Andrew J Burghardt 1 , Jin Long 2 , Tandy Aye 2 , Kyla Kent 2 , Ariana Strickland 2 , Mary B Leonard 2
  1. University of California San Francisco, San Francisco, CA, United States
  2. Stanford University, Stanford, CA, USA

Micro-finite element analysis (µFEA) of HR-pQCT images typically assumes homogenous material properties. This approach does not account for variability in tissue density, which is high during periods of rapid growth. Alternatively, material properties can be assigned proportional to each element’s BMD. Our objective was to evaluate the contribution of local density by comparing stiffness derived from homogenous and density-scaled µFEA in pediatric HR-pQCT.

The density-modulus relation was fit to a power function:

E(x,y,z)=Et*[BMD(x,y,z)/BMDt]γ

HR-pQCT scans from a cohort of young adults (n=42;age:21-30) were assumed to approximate homogenous properties (Fig1a). Therefore, in this population, optimization of the power equation coefficients should minimize the RMS difference between stiffness derived by scaled and homogenous models. We assumed the maximal density of fully mineralized bone (BMDt) to be 1200mg/cm3. To investigate how the optimized density-modulus function impacts µFEA outcomes in pediatric HR-pQCT, we applied this model to tibia and radius scans from a pediatric cohort (n=214;age=5-20). Differences were calculated by tanner stage. Regression analysis was used to evaluate the association with cortical bone features.

In young adults, minimization of RMSD of homogenous and density-scaled µFEA stiffness yielded power coefficient γ=2.25 and maximal modulus Et=15,974MPa (RMSD=7.9%;Fig1b). In the pediatric cohort, homogenous µFEA stiffness was, on average, significantly higher than density-scaled stiffness (RMSD=13.8%,p<0.01,Fig1c). Differences tended to increase by Tanner stage and were negatively associated with Ct.TMD (tibia:r2=0.48;radius:r2=0.73;p<0.0001).

In pediatric subjects, numerically higher µFEA-derived stiffness computed using homogenous tissue property assumptions likely reflects overestimation of the material properties themselves as well as a sub-optimal fine-structure segmentation that systematically overestimates bone volume. This differential increases with maturation. By accounting for tissue heterogeneity and providing threshold independence, density-scaled material properties likely improve the accuracy of µFEA for pediatric HR-pQCT, particularly across maturational stages and in the context of mineralization abnormalities.

629952db0b91b-ANZBMS_2022_DensityScaledFEA_FIG1.jpg