What question did this study set out to answer?

The research aims to create a system capable of real-time monitoring and visualization of dose distributions in proton therapy.

April 24, 2026Nuclear Techniques0 citationsOpen Access

Real-time reconstruction and visualization system for 3D dose distribution in particle radiotherapy based on GPU acceleration

GPGuo PengSichuan University CGChenyang GongSichuan University KZKeyao ZengFujian University of Technology

Key Points

The research aims to create a system capable of real-time monitoring and visualization of dose distributions in proton therapy.
Developed a GPU-accelerated system using CUDA for efficient dose reconstruction.
Integrated live data from ionization chambers and CT images.
Implemented a Python framework for data fusion and visualization of dose information.
The system reconstructs and visualizes dose distributions in 103.7 seconds, faster than the proton delivery time.
For a clinical plan, it accomplishes dose reconstruction in 82.33 seconds, demonstrating real-time capabilities.

Abstract

BackgroundIn particle therapy, real-time quality control and monitoring of three-dimensional (3D) dose distribution remains a significant technical challenge.PurposeThis study aims to develop a Graphics Processing Unit (GPU)-accelerated system for real-time 3D dose reconstruction and visualization of proton intensity-modulated radiotherapy.MethodsThe live data streams from a dose ionization chamber, a position ionization chamber, and the scanning control system with patient Computed Tomography (CT) images, treatment-planning data, and a precomputed beamlet matrix dijk for various positioning-error scenarios were integrated in the system. By leveraging Compute Unified Device Architecture (CUDA) acceleration, dynamic superposition of dose distributions with high efficiency was achieved. The software was implemented in a Python framework that handled data fusion and provided an interactive graphical user interface, and supported multidimensional visualization outputs such as dose images and Dose Volume Histogram (DVH). Finally, a computer equipped with an Advanced Micro Devices, Inc., (AMD) Ryzen75800 Central Processing Unit (CPU) and an NVIDIA GeForce 3060Ti GPU is used to test a TG119 abdominal single-field proton Intensity Modulated Proton Therapy (IMPT) plan consisting of 16 layers and 1 672 beam spots.ResultsTest results show that the system completes dose reconstruction and visualization for this plan in 103.7 s—faster than the plan's estimated proton delivery time of 104.21 s. In addition, for a clinical nasopharyngeal dual-field proton IMPT plan with 1 345 beam spots, the system completes dose reconstruction and visualization in 82.33 s, which is also shorter than the estimated delivery time of 83.01 s, demonstrating both real-time capability and clinical applicability.ConclusionsThis system proposed in this study provides real-time decision support for in-treatment target dose verification, dynamic organ-at-risk assessment, and adaptive adjustment of irradiation parameters, offering technical assurance to enhance quality control and clinical safety in particle therapy.

Ask AI

Helpful

Bookmark

View Full Paper