Continuous pose optimization improves sparse-view CT trajectory planning
Gradient-Based Trajectory Optimisation over Continuous Poses for Sparse-View Cone-Beam CT
Computer Vision and Pattern Recognition
Summary
Taking detailed 3D X-ray pictures often needs careful planning about where to place the X-ray source around the object. This paper studies a new way to decide those positions as continuous flexible choices rather than fixed sets, allowing more precise scanning paths. The authors show this method works well on different scanner setups and in simulations, helping find defects not seen by traditional circular scans. They also tested it with real equipment but found no clear improvement there yet. Their approach directly includes physical limits and X-ray absorption in the planning.
What this means in practice
- •For medical imaging engineers: Optimize scan paths in CT machines to improve coverage and defect detection with fewer X-ray views, adapting to various scanner geometries.
- •For industrial ct operators: Plan efficient sparse-view scans for industrial parts inspections that incorporate object shape and scanner constraints, reducing scan time and radiation dose.
Authors
Linda-Sophie Schneider, Simon Wittl, Gabriel Herl, Andreas Maier
Abstract
Trajectory optimisation for cone-beam computed tomography (CT) determines which information sparse-view scans acquire. Fixed candidate pools prevent off-grid refinement and require new object-specific precomputation for each acquisition manifold. We make every source pose an individual continuous variable and move all poses jointly by gradient ascent on the scanner's kinematic manifold. The objective combines soft-Tuy plane coverage, continuous View Covariance Loss, and an analytic attenuation-aware ray-bundle penalty. The same optimiser handles circular, limited C-arm, two-axis, and freesphere parametrisations. On a Defrise flange, continuous selection recovers laminar defects invisible to a circular orbit, matches discrete swap search on the free sphere at the sparser budget, and leads at the denser one, with the same objective evaluated in every arm. A moderate elevation band already recovers most of the free-sphere gain at the defects, so the same optimiser transfers to bounded scanner envelopes. Photon noise preserves the ordering on the flange and compresses it on a dense fuel nozzle. Sparseprescan planning benefits from matching prescan and planned acquisition manifolds. Selection takes seconds rather than minutes without an object-specific reconstruction basis. Prescan-planned poses were executed on a robot CT bench and reconstructed in a common frame, demonstrating feasibility but no consistent metric gain over uniform band sampling. Continuous pose optimisation incorporates attenuation and scanner constraints directly into sparse-view acquisition design.