Reconstructing a 3D Medical Image from a Few 2D Projections Using a B-Spline-Based Deformable Transformation

(1) Background: There was a need for 3D image reconstruction from series of 2D projections in medical applications. However, additional exposure to X-ray may harm human health. To alleviate it, minimizing the projection numbers is solution reduce exposure, but this would cause significant noise and artifacts. (2) Purpose: In study, method proposed minimal set using B-spline-based deformable transformation. (3) Methods: The inputs were which acquired previous treatment used as prior during current treatment. goal reconstruct new two inputs. deformed via displacement matrixes that interpolated by B-spline coefficients. coefficients solved with objective function, defined mean square error between reconstructed ground-truth projections. optimization process gradient function calculated, then updated. For acceleration purpose, computation reconstructions interpolation implemented on graphics processing unit (GPU). (4) Results: When scan angles more than 60°, quality significantly improved, comparable image. As less 30°, degraded. influence orientation minor. With application GPU acceleration, efficiency improved hundred times compared conventional CPU. (5) Conclusions: able generate high-quality few projections, amount ~ 20% total required standard introduction B-spline-interpolated matrix effective suppressing This could imaging time radiation patients under