Quantitative Underestimation of Two-Dimensional Anterior and Posterior Glenoid Bone Loss Measurements Varies According to Bone Morphology.

Abstract

Quantitative means of characterizing glenoid bone loss (GBL) remain a growing field of interest. Here, a 3D volumetric GBL estimation is compared against standard 2D methods for GBL computation to elucidate the effect of morphological differences in glenoid anatomy as a culprit for mis-estimating GBL. Twenty-two shoulder MRIs from healthy participants (19 ± 0.5 years) were segmented to isolate the glenoid. Anterior and posterior bone defects (5%-25%) were simulated according to established patterns of GBL in shoulder instability. Circle-of-best-fit and surface-based measurements of percent GBL were compared against the proposed 3D volumetric computation using Spearman correlations. Discrepancies in 2D measurements relative to 3D parameters were analyzed in the context of glenoid bony morphology derived from depth-based projection mapping. Very weak to moderate agreement was documented for both anterior and posterior defects. Direct comparison of 2D perfect-circle and surface-area methods against 3D measurements showed that 2D methods often over- or under-estimate percent GBL, which varies according to bone morphology. Greater variability in 3D volumetric GBL measurements was observed across all defects, suggesting greater morphological variability captured by the 3D-based computation. MRI-based volumetric analysis of 3D glenoid reconstruction may provide relevant insight for understanding the role of morphological differences in estimating critical bone loss toward improving clinical decision making for surgical management of shoulder instability.