The aim of this study was to propose a novel method for reconstructing the external body envelope from the low dose biplanar X-rays of a person. The 3D body envelope was obtained by deforming a template to match the surface profiles in two X-rays images in three successive steps: global morphing to adopt the position of a person and scale the template׳s body segments, followed by a gross deformation and a fine deformation using two sets of pre-defined control points. To evaluate the method, a biplanar X-ray acquisition was obtained from head to foot for 12 volunteers in a standing posture. Up to 172 radio-opaque skin markers were attached to the body surface and used as reference positions. Each envelope was reconstructed three times by three operators. Results showed a bias lower than 7mm and a confidence interval (95%) of reproducibility lower than 6mm for all body parts, comparable to other existing methods matching a template onto stereographic photographs. The proposed method offers the possibility of reconstructing body shape in addition to the skeleton using a low dose biplanar X-rays system.

This work underlines the potential of low-dose biplanar x-ray devices to make subject-specific FEMs for prediction of vertebral strength.

Finite element models (FEM) derived from qCT-scans were developed as a clinical tool to evaluate vertebral strength. However, the high dose, time and cost of qCT-scanner are limitations for routine osteoporotic diagnosis. A new approach considers using bi-planar dual energy (BP2E) X-rays absorptiometry to build vertebral FEM using synchronized sagittal and frontal plane radiographs. The purpose of this study was to compare the performance of the areal bone mineral density (aBMD) measured from DXA, qCT-based FEM and BP2E-based FEM in predicting experimental vertebral strength. Twenty eight vertebrae from eleven lumbar spine segments were imaged with qCT, DXA and BP2E X-rays before destructively tested in anterior compression. FEM were built based on qCT and BP2E images for each vertebra. Subject-specific FEM were built based on 1) the BP2E images using 3D reconstruction and volumetric BMD distribution estimation and 2) the qCT scans using slice by slice segmentation and voxel based calibration. Linear regression analysis was performed to find the best predictor for experimental vertebral strength (F); aBMD, modeled vertebral strength and vertebral stiffness. Areal BMD was moderately correlated with F (R = 0.74). FEM calculations of vertebral strength were highly to strongly correlated with F (R = 0.84, p < 0.001 for BP2E model and R = 0.95, p < 0.001 for qCT model). The results of this study suggest that aBMD accounted for only 74% of F variability while FE models accounted for at least 84%. For anterior compressive loading on isolated vertebral bodies, simplistic loading condition aimed to replicate anterior wedge fractures, both FEM were good predictors of F. Therefore FEM based on BP2E X-rays absorptiometry could be a good alternative to replace qCT-based models in the prediction of vertebral strength. However future work should investigate the performance of the BP2E-based model in vivo in discriminating patients with and without vertebral fracture in a prospective study.