Premium
SU‐D‐207‐05: Real‐Time Intrafractional Motion Tracking During VMAT Delivery Using a Conventional Elekta CBCT System
Author(s) -
Park YangKyun,
Sharp Gregory C.,
Gierga David P.,
Ye SungJoon,
Winey Brian A.
Publication year - 2015
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4923906
Subject(s) - imaging phantom , fiducial marker , cone beam computed tomography , tracking (education) , projection (relational algebra) , computer science , computer vision , nuclear medicine , dicom , match moving , dosimetry , image quality , artificial intelligence , medicine , motion (physics) , computed tomography , image (mathematics) , radiology , algorithm , psychology , pedagogy
Purpose: Real‐time kV projection streaming capability has become recently available for Elekta XVI version 5.0. This study aims to investigate the feasibility and accuracy of real‐time fiducial marker tracking during CBCT acquisition with or without simultaneous VMAT delivery using a conventional Elekta linear accelerator. Methods: A client computer was connected to an on‐board kV imaging system computer, and receives and processes projection images immediately after image acquisition. In‐house marker tracking software based on FFT normalized cross‐correlation was developed and installed in the client computer. Three gold fiducial markers with 3 mm length were implanted in a pelvis‐shaped phantom with 36 cm width. The phantom was placed on a programmable motion platform oscillating in anterior‐posterior and superior‐inferior directions simultaneously. The marker motion was tracked in real‐time for (1) a kV‐only CBCT scan with treatment beam off and (2) a kV CBCT scan during a 6‐MV VMAT delivery. The exposure parameters per projection were 120 kVp and 1.6 mAs. Tracking accuracy was assessed by comparing superior‐inferior positions between the programmed and tracked trajectories. Results: The projection images were successfully transferred to the client computer at a frequency of about 5 Hz. In the kV‐only scan, highly accurate marker tracking was achieved over the entire range of cone‐beam projection angles (detection rate / tracking error were 100.0% / 0.6±0.5 mm). In the kV‐VMAT scan, MV‐scatter degraded image quality, particularly for lateral projections passing through the thickest part of the phantom (kV source angle ranging 70°‐110° and 250°‐290°), resulting in a reduced detection rate (90.5%). If the lateral projections are excluded, tracking performance was comparable to the kV‐only case (detection rate / tracking error were 100.0% / 0.8±0.5 mm). Conclusion: Our phantom study demonstrated a promising Result for real‐time motion tracking using a conventional Elekta linear accelerator. MV‐scatter suppression is needed to improve tracking accuracy during MV delivery. This research is funded by Motion Management Research Grant from Elekta.