X-ray CT adaptation based on a 2D-3D deformable image registration framework using simulated in-room proton...

Area of Science:

• Medical Physics
• Radiation Oncology
• Image Processing

Background:

• Adaptive radiation therapy (ART) requires accurate patient positioning and anatomical tracking.
• Deformable image registration (DIR) is crucial for compensating anatomical changes during treatment.
• Proton radiography (pRads) offers potential for in-room imaging in proton therapy.

Purpose of the Study:

• To investigate the use of 2D-3D DIR with in-room proton radiography for compensating rigid and non-rigid transformations in clinical scenarios.
• To evaluate the feasibility of using low-dose pRads for adaptive proton therapy.
• To develop and assess a stopping criterion for DIR in the absence of ground truth proton CT.

Main Methods:

• Simulated Monte Carlo proton radiographies from head and neck/brain CT datasets.
• Application of realistic rigid and non-rigid deformations.
• Evaluation of integration-mode and list-mode detectors with varying proton statistics and fields of view (FOVs).
• Implementation and testing of a novel stopping criterion for DIR.

Main Results:

• Rigid and non-rigid transformations were effectively compensated using a limited number of low-dose pRads.
• 2D-3D DIR achieved accuracy comparable to 3D-3D DIR using 10 integration-mode or 2 list-mode pRads.
• The proposed stopping criterion effectively indicated optimal DIR convergence.
• Proton statistics and FOV size had minimal impact on accuracy, except for 2 integration-mode pRads.

Conclusions:

• In-room proton radiography combined with 2D-3D DIR is a promising approach for ART in proton therapy.
• The method demonstrates potential for accurate compensation of anatomical variations.
• Further computational optimization is needed for clinical validation and implementation.