EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An Integrated GPU-Accelerated Monte Carlo and AI-Based System for Real-Time Respiratory Motion Management in Thoracic Proton Therapy

Hadi TALESHİ AHANGARİ, Ahmad SAUD DAKHEL ALKATAN

Advances in Medical Pyhsics and Applied Sciences · 2026

DOI: 10.63187/ampas.49 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Advances in Medical Pyhsics and Applied Sciences, 2026. Verfasst von Hadi TALESHİ AHANGARİ und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Purpose

Respiratory motion is a major challenge in proton therapy (PT) for thoracic tumors, causing interplay effects that degrade dose distribution. While Monte Carlo (MC) simulation is the gold standard for dose calculation, its computational burden hinders real-time adaptive use. This study aims to develop and validate an integrated system combining a Graphics Processing Unit (GPU)-accelerated MC engine with Artificial Intelligence (AI) models for real-time motion management in thoracic PT.

Methods

A three-module software platform was developed: 1) a CUDA/C++ based GPU-MC engine simulating proton transport with variance reduction techniques; 2) an LSTM network trained on 4D-CT data to predict respiratory motion; and 3) a 3D U-Net for auto-segmentation of organs-at-risk. The system was evaluated on 50 patient scenarios against static plans and conventional gating using gamma analysis, RMSE, Dice scores, and dosimetric indices (V95%, mean OAR doses).

Results

The GPU-MC engine achieved a ~285x speedup versus single-core CPU (simulation time 99.2% gamma passing rate versus Geant4. The LSTM predicted motion with

Zitieren

APA

AHANGARİ, H. T., & ALKATAN, A. S. D. (2026). An Integrated GPU-Accelerated Monte Carlo and AI-Based System for Real-Time Respiratory Motion Management in Thoracic Proton Therapy. Advances in Medical Pyhsics and Applied Sciences. https://doi.org/10.63187/ampas.49

Vancouver

AHANGARİ HT, ALKATAN ASD. An Integrated GPU-Accelerated Monte Carlo and AI-Based System for Real-Time Respiratory Motion Management in Thoracic Proton Therapy. Advances in Medical Pyhsics and Applied Sciences. 2026. doi:10.63187/ampas.49

BibTeX
@article{AHANGAR2026,
  author = {AHANGARİ, Hadi TALESHİ and ALKATAN, Ahmad SAUD DAKHEL},
  title = {An Integrated GPU-Accelerated Monte Carlo and AI-Based System for Real-Time Respiratory Motion Management in Thoracic Proton Therapy},
  journal = {Advances in Medical Pyhsics and Applied Sciences},
  year = {2026},
  doi = {10.63187/ampas.49}
}
Kostenlos starten →