Author: Doris Dimitriadis/Dimitriadou, M. Saiful Huq, Ronald John Lalonde, Fang Li, Noor Mail, Adam Olson 👨🔬
Affiliation: UPMC Hillman Cancer Center, UPMC Hillman Cancer Center and University of Pittsburgh School of Medicine, UPMC 🌍
Purpose: The objective of this study was to qualitatively evaluate Prostate Hypersight Cone-Beam CT (HSCBCT) images and assess its capability for Limbus auto-contouring. The qualitative evaluation and assessment of Limbus contours by radiation oncologists is reported in this work.
Methods: Eleven prostate cancer patients aged 60–75 underwent initial CT simulation (CT sim) scans, followed by HSCBCT scans acquired on the Halcyon system (Varian Medical Systems, Palo Alto, CA) during treatment. HSCBCT and CT sim images were exported to Limbus, where prostate structure sets were selected and imported into Eclipse with Limbus-generated contours. Radiation oncologists reviewed and rated each organ-at-risk (OAR) and pelvic node contour on both HSCBCT and CT sim images. The scoring criteria were: 1 = no changes needed, 2 = minor deviation acceptable, 3 = significant deviation requiring change, and 4 = not applicable. Additionally, a quantitative comparison of bony anatomy volumes was conducted to assess reconstruction accuracy in HSCBCT images.
Results: Limbus auto-contours for the rectum, cauda equina, penile bulb, sigmoid colon, and bony structures received scores of 1, indicating no adjustments were needed. For bladder contours, nine patients required no adjustment on CT sim and seven on HSCBCT. CT sim marginally outperformed HSCBCT in lymph node and seminal vesicle contours under the "no adjustment needed" criterion, whereas HSCBCT showed slight advantages under the "minor deviation acceptable" category. Volume differences for the femoral head and pubic bone were 0.234 ± 1.98% and 4.87 ± 1.86%, respectively.
Conclusion: High-quality HSCBCT images obtained directly in the treatment room facilitated Limbus contouring, yielding results closely comparable to CT sim. The qualitative evaluation of HSCBCT images and Limbus contouring data supports its potential in adaptive radiotherapy applications.