Monte Carlo dose calculation algorithm on a distributed system

Stéphane Chauvie, Matteo Dominoni, Piergiorgio Marini, Michele Stasi, Maria Grazia Pia, Giuseppe Scielzo

Research output: Contribution to journalArticlepeer-review


The main goal of modern radiotherapy, such as 3D conformal radiotherapy and intensity-modulated radiotherapy is to deliver a high dose to the target volume sparing the surrounding healthy tissue. The accuracy of dose calculation in a treatment planning system is therefore a critic issue. Among many algorithms developed over the last years, those based on Monte Carlo proven to be very promising in terms of accuracy. The most severe obstacle in application to clinical practice is the high time necessary for calculations. We have studied a high performance network of Personal Computer as a realistic alternative to a high-costs dedicated parallel hardware to be used routinely as instruments of evaluation of treatment plans. We set-up a Beowulf Cluster, configured with 4 nodes connected with low-cost network and installed MC code Geant4 to describe our irradiation facility. The MC, once parallelised, was run on the Beowulf Cluster. The first run of the full simulation showed that the time required for calculation decreased linearly increasing the number of distributed processes. The good scalability trend allows both statistically significant accuracy and good time performances. The scalability of the Beowulf Cluster system offers a new instrument for dose calculation that could be applied in clinical practice. These would be a good support particularly in high challenging prescription that needs good calculation accuracy in zones of high dose gradient and great dishomogeneities.

Original languageEnglish
Pages (from-to)159-163
Number of pages5
JournalNuclear Physics B - Proceedings Supplements
Publication statusPublished - Sep 2003

ASJC Scopus subject areas

  • Nuclear and High Energy Physics


Dive into the research topics of 'Monte Carlo dose calculation algorithm on a distributed system'. Together they form a unique fingerprint.

Cite this