X2BR: High-fidelity 3D bone reconstruction from a planar X-ray image with hybrid neural implicit methods

Accurate 3D bone reconstructions are critical for surgical planning, implant design, and tracking musculoskeletal disorders. 3D bone reconstruction from a single planar X-ray remains challenging due to anatomical complexity and limited input. We propose X2BR, a hybrid neural implicit framework combining continuous volumetric reconstruction (X2B) with template-guided non-rigid registration (R). X2B employs a ConvNeXt-based encoder to extract spatial features from X-rays and predict high-fidelity 3D bone occupancy fields without relying on statistical shape models. To refine anatomical accuracy, X2BR integrates a patient-specific template mesh, constructed using YOLOv9-based detection and the SKEL biomechanical skeleton model. The coarse reconstruction is aligned to the template using geodesic-based coherent point drift, enabling anatomically consistent 3D bone volumes. Experimental results show that X2B achieves the highest numerical accuracy, with an IoU of 0.952 and Chamfer-L1 distance of 0.005, outperforming recent baselines including X2V and D2IM-Net. Building on this, X2BR incorporates anatomical priors via YOLOv9-based detection and biomechanical template alignment, leading to reconstructions that offer superior anatomical realism, especially in rib curvature and vertebral alignment, while slightly lower in IoU (0.875). This accuracy vs. visual consistency trade-off between X2B and X2BR highlights the value of hybrid frameworks for clinically relevant 3D reconstructions.