Precomputed compressed projection matrices cut reconstruction time and memory latency for faster, more accurate 3D x-ray tomography.
A preliminary reconstruction guides parameter updates to match scattering conditions, improving medical image quality and reducing artifacts.
A 3D CNN trained on simulated digital phantoms cuts CBCT cone-beam artifacts while avoiding large datasets and heavy reconstruction.
Shim stabilization, motion sensors, and calibration enable manual C-arm imaging to produce accurate intraprocedural 3D reconstructions.