Sparse deconvolution of proton radiography data to estimate water equivalent thickness maps

Area of Science:

• Medical Physics
• Radiation Oncology
• Imaging Science

Background:

• Proton therapy planning relies on converting CT scans to stopping powers, which introduces uncertainties affecting dose conformity.
• Proton radiography offers direct measurement of proton energy loss but is limited by blurred depth-dose profiles.

Purpose of the Study:

• To develop an iterative algorithm for extracting high-resolution water equivalent thickness (WET) maps from proton radiography data.
• To improve the accuracy and robustness of WET map estimation in proton therapy.

Main Methods:

• Implemented an iterative deconvolution algorithm using depth-dose profiles from a multilayer ionization chamber.
• The algorithm assumes depth-dose curves are a function of WET and utilizes sparse representation.
• A variant integrates planning CT data for enhanced deconvolution accuracy.

Main Results:

• The proposed method, even without CT priors, surpasses existing techniques in accuracy.
• Integration of CT data further refines results, achieving 1.5 mm accuracy in WET estimation.
• The algorithm demonstrated effectiveness on both synthetic data and phantom acquisitions.

Conclusions:

• The deconvolution algorithm significantly enhances WET map accuracy from proton radiography.
• The method proves robust against setup errors and anatomical variations.
• Proton radiography with range probes offers a simple yet accurate approach to reduce range uncertainty in proton therapy.