Acuros XB increased synthetic CT dose differences but remained clinically acceptable

KEY POINTS

• The study included 39 prostate cancer patients and 17 glioma patients treated using MRI-only VMAT workflows. Prostate plans prescribed 78 Gy in 39 fractions, while most glioma plans prescribed 60 Gy in 30 fractions.
• Plans were recalculated on conventional CT and synthetic CT using the convolution-superposition Anisotropic Analytical Algorithm and the linear Boltzmann transport equation-based Acuros XB algorithm. Acuros XB was evaluated using both dose-to-medium and dose-to-water reporting.
• In prostate plans, mean synthetic CT-to-CT target differences were no greater than 0.5% with AAA and 1.4% with Acuros XB. For PTV D95%, differences were 0.5% with AAA, 1.0% with Acuros dose-to-medium, and 0.8% with dose-to-water.
• In glioma plans, mean target-dose differences were no greater than 0.5% with AAA and 0.8% with Acuros XB. PTV D98% differed by 0.5%, 0.8%, and 0.8% with AAA, Acuros dose-to-medium, and Acuros dose-to-water, respectively.
• Prostate organ-at-risk differences remained small: mean volume-based differences were at most 0.4 percentage points, and dose-based differences were at most approximately 0.35 Gy. Glioma organ-at-risk differences were at most 0.7 Gy and showed no consistent direction or statistically significant change.
• Acuros XB produced approximately twofold larger synthetic CT-to-CT differences than AAA for several metrics. Dose-to-medium was generally the most sensitive mode, particularly in bone, although the absolute deviations remained within accepted clinical tolerances.
• Atlas-based and deep-learning synthetic CT methods produced closely comparable results, generally differing by no more than 0.2 percentage points. Limitations included a single software platform, one treatment-planning system, rigid CT–MRI registration, and evaluation limited to prostate and brain VMAT.

CLINICAL TAKEAWAY

SOURCE

Physics and Imaging in Radiation Oncology