Researchers have developed an innovative approach that combines Monte Carlo simulation with deep learning to improve the speed and accuracy of quality assurance in radiation therapy. This method tackles a major hurdle in ensuring precise dose verification during treatments.
Addressing Challenges in Dose Verification
Electronic portal imaging devices (EPID) play a vital role in real-time, in vivo dose verification for radiation therapy. Monte Carlo (MC) simulations serve as the gold standard for dose calculations, but they present a trade-off: higher particle counts yield greater accuracy yet demand extensive computation time, while fewer particles produce noisy results that undermine reliability.
The MC-DL Integration
To overcome this, experts led by Professor Fu Jin integrated a GPU-accelerated MC code called ARCHER with the SUNet neural network, designed specifically for denoising tasks. In tests using lung cancer intensity-modulated radiation therapy (IMRT) cases, the team generated noisy EPID transmission dose data at varying particle levels: 1×106, 1×107, 1×108, and 1×109. They trained SUNet to refine low-particle data, using the high-fidelity 1×109 dataset as the reference standard.
Impressive Performance Gains
The combined MC-deep learning (MC-DL) framework delivers outstanding results in both efficiency and precision. For the noisy 1×106-particle data, SUNet denoising boosted the structural similarity index (SSIM) from 0.61 to 0.95 and raised the gamma passing rate (GPR) from 48.47% to 89.10%. At the 1×107-particle level, which balances speed and accuracy, denoised outputs achieved an SSIM of 0.96 and a GPR of 94.35%. The 1×108-particle data reached a GPR of 99.55% post-processing.
The denoising process takes just 0.13 to 0.16 seconds, slashing total computation time to 1.88 seconds for 1×107 particles and 8.76 seconds for 1×108 particles. Processed images show reduced graininess and smooth dose profiles that preserve essential clinical details, making this technique highly practical for radiotherapy quality assurance.
Advancing Clinical Applications
This breakthrough holds significant promise for online adaptive radiotherapy (ART), where quick dose checks help reduce patient discomfort and account for anatomical changes. The approach offers flexibility: 1×107 particles suit time-critical situations, while 1×108 particles ensure superior precision in complex scenarios.
“By integrating the accuracy of Monte Carlo simulation with the computational efficiency of deep learning, we have developed a practical solution that addresses the critical clinical need for rapid and reliable patient-specific quality assurance,” stated Professor Fu Jin. “This technology not only enhances existing radiation therapy workflows but also establishes a foundation for advanced applications, such as 3D dose reconstruction and broader implementation across diverse anatomical sites.”
Future efforts will extend the model to additional treatment sites, refine the SUNet architecture, and investigate other neural networks to further enhance dose prediction.
