Sensitivity of Trabecular Mapping to Sliding Semilandmark Placement

Abstract

Analyses of bone microstructure based on single volumes of interest (VOIs) are limited in their ability to quantify variation in trabecular properties across a joint. Geometric morphometric methods can overcome these limitations by utilizing sliding semilandmarks to locate multiple VOIs within a trabecular structure. Variation in the placement of semilandmarks, however, may change collected bone parameter values. Here we quantify the effect of intraobserver error in delineating the articular surface on sliding semilandmark placement and sampled trabecular properties.Proximal tibiae of 12 individuals that are part of the Texas State University Donated Skeletal Collection were scanned using micro‐computed tomography at the Forensic Anthropology Center at Texas State. Original grey scale image stacks were converted to binary images, and surface models of the exterior bone surface were created. Medial condyles were trimmed from their full surface models three times over a period of weeks. Sliding semilandmarks (n=141) were distributed on the condyle surfaces and then slid to minimize the bending energy of the thin‐plate spline function relative to the updated Procrustes average. Each sliding semilandmark was then used to locate a single VOI just deep to the cortical shell. To investigate the effect of landmark placement on trabecular bone parameters, each VOI was moved 0.5–1.5 mm to one of 10 neighboring positions. Bone volume fraction (BV/TV) and trabecular thickness (Tb.Th) were calculated for each VOI at each of the 11 positions. VOIs that moved outside the trabecular structure were removed from further analyses. 10,000 randomized combinations of the 11 positions of each VOI were created and correlations between bootstrap samples were calculated.Sliding semilandmark placement differed by 0.27 mm on average. Landmarks most affected by intraobserver error were those placed along the edge of the condylar surface (mean = 0.38 mm), while those on the interior portion of the condyle were affected less (mean = 0.21 mm). Sampled BV/TV and Tb.Th vary with VOI position, although there is a strong correlation among the 11 sampled values for BV/TV (r=0.83) and Tb.Th (r=0.80). These results suggest that VOI positions calculated from sliding semilandmarks provide consistent information about bone microstructure even when VOI placement is forced to deviate beyond average variation in sliding landmark position.Support or Funding InformationThis work was supported in part by instrumentation funded by the NSF under grant NSF: MRI 1338044.This abstract is from the Experimental Biology 2019 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.