Comparing methods to improve cone-beam computed tomography for dose calculations in adaptive proton therapy

Area of Science:

• Medical Physics
• Radiotherapy
• Image Processing

Background:

• Proton therapy dose monitoring relies on repeated CT scans, which may not reflect real-time anatomy.
• In-room cone-beam CT (CBCT) has lower image quality, limiting proton dose calculation accuracy.
• Adaptive proton therapy requires high-quality imaging for accurate dose monitoring.

Purpose of the Study:

• To evaluate different quality-improvement methods for CBCTs (synthetic CTs; sCTs).
• To assess the suitability of sCTs for adaptive proton therapy in head-and-neck cancer patients.
• To compare the accuracy of sCTs against repeated CT scans for proton dose calculations.

Main Methods:

• Four sCT generation methods were assessed: intensity-correction, two deformable image registration, and deep learning-based.
• Thirty-five CBCTs from 24 head-and-neck cancer patients were used.
• sCTs were evaluated for CT number accuracy, proton range accuracy, and dose recalculation accuracy using DVH parameters compared to reCTs.

Main Results:

• All four sCT methods improved CBCT image quality while preserving anatomy.
• Proton range differences between sCT methods were small, often less than sCT vs. reCT differences (1.0-1.1 mm).
• DVH parameter differences were generally small between sCT methods, with outliers often consistent across methods.

Conclusions:

• All evaluated sCT methods enable accurate proton dose calculation.
• sCTs effectively preserve anatomical information for adaptive proton therapy.
• These sCT methods are valuable tools for adaptive proton therapy in head-and-neck cancer.