EUReCA: End-to-End Unified CBCT Reconstruction Across Clinical Acquisitions

EUReCA is a geometry‑aware, feed‑forward model that reconstructs cone‑beam CT images from 8–100 projections without needing view‑specific retraining. It encodes each projection with a shared multiscale backbone, back‑projects features into 3‑D grids, aggregates variable views at fixed cost, and refines the volume using a global transformer and decoder. On simulated and real patient data, EUReCA outperforms existing feed‑forward and iterative methods, achieving high PSNR and superior performance across all tested view counts.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

• llms