S2060
Physics - Image acquisition and processing
ESTRO 2026
Purpose/Objective: The Gamma Knife CBCT system is essential for precise patient positioning, for instance in fractionated treatments. However, the geometry of the imaging system and its circular arc scan trajectory do not respect Tuy’s condition for data sufficiency, resulting in significant cone-beam artifacts, which increase along the patient’s superior direction. As improving image quality for on-board imaging is relevant for better real- time visualization and image registration, the objective is to investigate new scan trajectories that could enhance image quality, given the machine’s geometrical and physical constraints. Material/Methods: To reduce cone-beam artifacts, various scan trajectories added to the current 200-degree arc are investigated, such as lines, helices and arcs higher in the superior direction. The sampling completeness for these scan trajectories is quantified through realistic tomographic incompleteness maps [1,2], considering discretized trajectories, finite detector size and overall CBCT geometry of the Gamma Knife. The CBCT system is simulated from a CT reconstruction of a head phantom and a virtual phantom with pairs of parallel disks along the central axis. A set of projections is first generated from the input volume through polychromatic forward projection with added noise and Monte Carlo-simulated scatter. The volume is then reconstructed with an iterative reconstruction algorithm based on weighted least-squares, which supports any trajectory and has more potential for further image quality improvements than the filtered back-projection algorithm currently employed in the clinic. Both simulations and reconstructions are performed using tools developed at Elekta. Results: Figure 1 shows that the line-arc-line trajectory achieves the greatest improvement in sampling completeness within the reconstructed volume. Despite requiring fewest additional projections (100) compared to a single arc, this trajectory reaches incompleteness values of (0.01±0.01) (mean±SD), as opposed to (0.03±0.02), (0.06±0.05) and (0.11±0.07) for the arc-line-arc, helix-arc-helix and single arc, respectively. Visual inspections of the reconstructed head and disk phantoms confirmed that the line-arc- line trajectory exhibits the greatest reduction in cone- beam artifacts. Figure 2 shows images reconstructed from the line-arc-line trajectory, compared to those from the current arc trajectory. With this new scan trajectory, the prominent cone-beam artifacts in the skull are mostly gone, and the disk pairs inside the virtual phantom are visually resolved at all axial positions.
Conclusion: Theoretical calculations and simulations have shown that the line-arc-line scan trajectory is promising for removing cone-beam artifacts in Gamma Knife CBCT images. The next phase is to conduct experiments on a research Gamma Knife to investigate the line-arc-line trajectory. References: [1] R. Clackdoyle and F. Noo, “Quantification of Tomographic Incompleteness in Cone-Beam
Made with FlippingBook - Share PDF online