298 - 3D Surface Cherenkov Mapping for Quantitative Beam Deviation Assessment and Cumulative Cherenkov Image Fusion in Non-Coplanar RT

Purpose/Objective(s):

Cherenkov imaging provides direct visualization of dose delivery and is used clinically to confirm that a treatment is delivered as planned. During RT treatments, images are captured by two ceiling-mounted cameras, and displayed with real-time Cherenkov-background video overlay. While 2D views are sufficient for therapists and clinicians to identify gross errors, spatial beam measurements require emission to be localized in 3D space. Here, we develop a framework to spatially locate Cherenkov emission by texturing the 3D CT-based reference surface meshes with Cherenkov emission. We present spatial analysis of these 3D Cherenkov maps to measure absolute beam outline deviation on the target’s surface both in patient’s and beam frame of reference. Finally we demonstrate co-localization of clinical Cherenkov surface maps during a case of non-coplanar intensity modulated radiotherapy.

Materials/Methods:

A raycasting algorithm was developed to project 2D cumulative Cherenkov images onto interpolated surfaces generated from CT volumes. Raycasting accuracy was verified with a known 4x11 circle pattern on a flat phantom. Simulated beam vs. surface deviations were evaluated on a plastic phantom and a tissue-mimicking anthropomorphic phantom, irradiated with A 10x10 cm² 6X RT beam offset at incremental 1 cm and 1 mm shifts in the horizontal plane. Iterative Closest Point (ICP) metrics were used on beam’s eye view transformed Cherenkov maps to measure beam deviations. Clinical patient Cherenkov image and CT datasets with intrafraction couch angle changes were obtained under an IRB-approved retrospective clinical trial and raycasting was used to generate a cumulative Cherenkov image of non-coplanar fields at a fixed couch angle of zero degrees.

Results:

2D images with 0.5mm spatial resolution were projected onto an interpolated CT-based surface map with 0.6 mm slice thickness. Projection spatial accuracy of 0.85 mm ± 0.15 mm in X and 0.14 mm ± 0.35 mm in Y was measured on a known flat board pattern phantom. ICP metrics achieved an accuracy of 1.5 mm or better in both X and Y for the flat phantom. On the Annie phantom where anatomical occlusions prevented complete imaging of the 10x10 cm² treatment field, ICP metrics still achieved an accuracy of 2 mm or better in the X and Y planes. Finally, 4 Cherenkov fields with couch angles of 0,5,15, and 355 degrees were raycast onto their respective CT surfaces and transformed to a couch angle of 0 degrees to produce an cumulative image.

Conclusion:

For the first time, quantitative spatial evaluation of Cherenkov images was achieved by fusing 2D Cherenkov imaging with 3D CT datasets. This fusion enabled spatial comparison of multiple Cherenkov fields in patients with intrafraction couch kicks. Clinical implementation of spatially calibrated Cherenkov imaging has the potential to quantify day to day beam deviations and enhance visualization during VMAT treatments, particularly in cases where gantry occlusions partially or fully obstruct one camera’s viewpoint.