Novel radiation therapy QA method: Monte Carlo simulation meets deep learning for fast, accurate epid transmission dose generation

(Press-News.org) Bridging Speed and Accuracy in Radiation Therapy QA
Led by Professor Fu Jin, the study addresses a critical challenge in radiation therapy: balancing the computational speed and accuracy of EPID-based dose verification. EPID has emerged as a key tool for real-time in vivo dose verification. However, MC simulation—long regarded as the "gold standard" for dose calculation—faces a dilemma: increasing the number of simulated particles ensures higher accuracy but at the cost of significantly longer computation times, whereas reducing the particle count introduces disruptive noise that compromises result reliability.

Integrated MC-DL Technology
To address this challenge, the team combined the GPU-accelerated MC code ARCHER with the SUNet neural network—a sophisticated deep learning architecture specialized in denoising. Using lung cancer IMRT cases, they first generated noisy EPID transmission dose data with four different particle numbers (1×10⁶, 1×10⁷, 1×10⁸, 1×10⁹) via ARCHER. SUNet was then trained to denoise the low‑particle‑number data, with the high‑fidelity 1×10⁹ particle dataset serving as the gold‑standard reference for supervision.

Remarkable Outcomes: Speed and Accuracy Achieved
The integrated MC‑DL framework demonstrated exceptional performance in both computational speed and dosimetric accuracy. When processing the originally noisy 1×10⁶‑particle data, SUNet denoising improved the structural similarity index (SSIM) from 0.61 to 0.95 and increased the gamma passing rate (GPR) from 48.47% to 89.10%. For the 1×10⁷‑particle dataset—representing an optimal trade‑off—the denoised results achieved an SSIM of 0.96 and a GPR of 94.35%, while the 1×10⁸‑particle case reached a GPR of 99.55% after processing. The denoising step itself required only 0.13–0.16 seconds, reducing the total computation time to 1.88 s for the 1×10⁷‑particle level and to 8.76 s for the 1×10⁸‑particle level. The denoised images exhibited markedly reduced graininess, with smooth dose profiles that retained clinically relevant features—confirming the practical viability of this approach for efficient QA in radiotherapy.

Empowering Clinical Practice and Future Research
This advancement is particularly impactful for online ART, where rapid dose verification is essential to minimize patient discomfort and mitigate anatomical variations during treatment. The method offers a flexible solution: 1×10⁷ particles strikes an optimal balance between speed and accuracy for time-sensitive scenarios, while 1×10⁸ particles provide higher precision for demanding cases.

"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" said 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."

The team plans to expand the model to other treatment sites, optimize the SUNet architecture further, and explore additional neural network approaches to refine dose prediction capabilities.

The complete study is via by DOI: https://doi.org/10.1007/s41365-026-01898-2

END