Joseph Webster Stayman PhD, Presenter: Research Grant, Varian Medical Systems, Inc

Adam S. Wang PhD, Abstract Co-Author: Research support, Siemens AG

Wojciech Zbijewski PhD, Abstract Co-Author: Research Grant, Carestream Health, Inc

Yoshito Otake, Abstract Co-Author: Research support, Siemens AG

Jeffrey H. Siewerdsen PhD, Abstract Co-Author: Research Grant, Siemens AG Consultant, Siemens AG Research Grant, Carestream Health, Inc Royalties, Elekta AB

Interventional cone-beam CT differs greatly from diagnostic CT not only in highly flexible positioning of the source and detector, but also in that interventional imaging tasks typically involve well-posed detection and localization of targets which have been identified in pre-operative 3D imaging and planning. We propose to leverage this wealth of patient- and task-specific prior knowledge to design customized source-detector trajectories for subsequent intraoperative CBCT acquisitions to maximize imaging task performance.

Task-based performance in 3D imaging is predictable using analytical models of the imaging chain. Task-based detectability index, for example, can be computed upon specification of a task function, acquisition geometry, trajectory, detector physics, reconstruction process; and the patient anatomy. Using a preoperative CT volume to integrate patient-dependence, we compute a marginal detectability index related to individual rotation angle/obliquity pairs of an interventional C-arm. A task-based trajectory is formed by successively finding the angle pair yielding the greatest detectability (e.g., the “next best view”) and adding it to a growing collection of angles. The trajectory design approach was applied to a simulated pulmonary nodule detection task where the data from a task-driven noncircular orbit was reconstructed using a model-based iterative approach.

The task-based trajectories designed for the pulmonary nodule detection task were largely continuous despite the lack of a continuity constraint and tended to avoid long radiological path lengths (e.g., avoiding projections involving overlap of the nodule with bone or a surgical tool). Image reconstructions using the task-based orbit show excellent visualization of the nodule. By comparison, the nodule was obscured in reconstructions from sub-optimal orbits due to noise/limited spatial resolution.

Leveraging patient-specific information and analytical model for task-based imaging performance within the 3D image acquisition process allowed the design of customized orbits that maximize task performance in image-guided interventions.

Task-based trajectories yield improved imaging performance over standard orbits and can potentially automatically overcome challenging imaging scenarios near high-density objects and bone.

Stayman, J, Wang, A, Zbijewski, W, Otake, Y, Siewerdsen, J, Imaging-task-Optimized, Source-detector Trajectory Design and Reconstruction in 3D Interventional Imaging.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13022619.html