Tumor Motion Effects on Target Definition with the 6-Second Hypersight Cone-Beam CT In Adaptive Radiotherapy
Abstract
Purpose
The HyperSight 6-second rapid scan has been used clinically for internal-target-volume(ITV) definition. However, limited temporal sampling may lead to ITV underestimation, particularly for irregular motion. Using a joint dynamic reconstruction and motion-estimation framework(DREME), we extracted time-resolved tumor motion and evaluated its impact on target definition.
Methods
DREME reconstructs a dynamic CBCT sequence from a conventional CBCT scan by estimating a reference CBCT and a motion model, which generates per-projection deformation-vector-fields(DVFs) to deform the reference into projection-specific CBCTs. Each patient has two 60-second CBCT scans from two treatment fractions, both processed by DREME. Tumors were contoured on the reference CBCT and propagated to per-projection CBCTs using DVFs. Each 60-second scan was subdivided into ten 6-second rapid scans to emulate the HyperSight protocol. For each rapid scan, the ITV was derived from the maximum-intensity-projection of tumor contours and expanded to a Planning-Target-Volume(PTV) using clinically-defined margins(3-5mm). Intra-scan analysis compared each rapid-scan PTV with ITVs from the other nine rapid scans to compute target coverage, defined as the fraction of ITV within the PTV. For each PTV, mean and minimum target coverage was calculated among all comparisons. Inter-fraction analysis compared between two CBCT fractions of each patient, with each fast-scan PTV from one fraction compared to the 10 ITVs from another fraction. The impact of PTV margin was evaluated.
Results
Across five patients, intra-scan analysis yields the mean(±s.d.) and minimum(±s.d.) target coverage of 99.18±1.08%, and 90.4±9.4%. Inter-fraction analysis shows mean and minimum coverage of 98.3±0.6% and 89.0±4.2%. PTV margins of 1–5mm were tested in intra-scan analysis, and some cases showed <95% minimum coverage despite a 5mm margin, demonstrating the potential need of motion-customized margin for fast HyperSight scans.
Conclusion
For HyperSight rapid scans, both intra-scan and inter-fraction analyses achieved overall high mean target coverage. However, tumors with irregular motion may be inadequately covered.