[Paper Review] Monte Carlo modeling photon-tissue interaction using on-demand cloud infrastructure

Ethan P. M. LaRochelle, P. Arce+1 more|arXiv (Cornell University)|May 3, 2020

Optical Imaging and Spectroscopy Techniques38 references4 citations

TL;DR

This paper presents a scalable, containerized Monte Carlo simulation framework for photon-tissue interactions using on-demand cloud infrastructure, enabling parallel execution of non-thread-safe and non-GPU-optimized applications. By deploying 154 simulations across 64 containers on 4 servers, the method achieved up to 97% reduction in execution time compared to sequential processing, demonstrating high efficiency for medical physics modeling in cloud environments.

ABSTRACT

Purpose: This work advances a Monte Carlo (MC) method to combine ionizing radiation physics with optical physics, in a manner which was implicitly designed for deployment with the most widely accessible parallelization and portability possible. Methods: The current work updates a previously developed optical propagation plugin for GEANT4 architecture for medically oriented simulations (GAMOS). Both virtual-machine (VM) and container based instances were validated using previously published scripts, and improvements in execution time using parallel simulations are demonstrated. A method to programmatically deploy multiple containers to achieve parallel execution using an on-demand cloud-based infrastructure is presented. Results: A container-based GAMOS deployment is demonstrated using a multi-layer tissue model and both optical and X-ray source inputs. As an example, the model was split into 154 simulations which were run simultaneously on 64 separate containers across 4 servers. Conclusions: The container-based model provides the ability to execute parallel simulations of applications which are not inherently thread-safe or GPU-optimized. In the current demonstration, this reduced the time by at most 97% compared to sequential execution. The code and examples are available through an interactive online interface through links at: https://sites.dartmouth.edu/optmed/research-projects/monte-carlo-software/

Motivation & Objective

• To enable efficient, scalable simulation of photon-tissue interactions in medical physics using on-demand cloud resources.
• To overcome limitations of non-thread-safe and non-GPU-optimized applications in high-performance computing environments.
• To develop a portable and portable deployment model using containers and virtual machines for GAMOS-based simulations.
• To demonstrate significant speedup in simulation time through parallel execution on cloud infrastructure.
• To provide open access to code and interactive examples for reproducibility and community use.

Proposed method

• The authors extended the GAMOS (GEANT4-based) platform with an optical propagation plugin to simulate both X-ray and optical photon interactions in multi-layered tissue models.
• They deployed containerized instances using Docker to ensure portability and isolation across heterogeneous cloud environments.
• Virtual machine and container-based instances were validated using previously published benchmark scripts to ensure simulation accuracy.
• A programmatic deployment pipeline was implemented to dynamically launch and manage multiple containers across multiple cloud servers.
• Parallel simulations were distributed across 64 containers on 4 servers, with workloads split across 154 individual simulation tasks.
• The cloud infrastructure enabled automatic scaling and load balancing for high-throughput Monte Carlo simulations.

Experimental results

Research questions

• RQ1 Can containerized, cloud-based deployment significantly accelerate Monte Carlo simulations of photon-tissue interactions?
• RQ2 How effective is the proposed method in reducing simulation time for non-thread-safe and non-GPU-optimized applications?
• RQ3 Can the framework maintain simulation accuracy while enabling high parallelization across distributed cloud resources?
• RQ4 What is the performance gain when scaling from sequential to parallel execution using on-demand cloud infrastructure?
• RQ5 How portable and reproducible is the simulation framework across different computing environments?

Key findings

• The container-based deployment achieved up to a 97% reduction in simulation time compared to sequential execution.
• The method successfully enabled parallel execution of non-thread-safe and non-GPU-optimized applications via containerization.
• Validation using previously published scripts confirmed the accuracy of the simulation results across both VM and container instances.
• The framework demonstrated scalability by distributing 154 simulations across 64 containers on 4 servers in a cloud environment.
• The code and interactive examples are publicly available through a web interface, enhancing reproducibility and community access.
• The approach enables efficient, portable, and high-throughput simulation of photon-tissue interactions in medical physics applications.