Demonstration of scan path optimization in proton therapy

Joanne H. Kang, Jan J. Wilkens, Uwe Oelfke

Research output: Contribution to journalArticlepeer-review

29 Scopus citations

Abstract

A three-dimensional (3D) intensity modulated proton therapy treatment plan to be delivered by magnetic scanning may comprise thousands of discrete beam positions. This research presents the minimization of the total scan path length by application of a fast simulated annealing (FSA) optimization algorithm. Treatment plans for clinical prostate and head and neck cases were sequenced for continuous raster scanning in two ways, and the resulting scan path lengths were compared: (1) A simple back-and-forth, top-to-bottom (zigzag) succession, and (2) an optimized path produced as a solution of the FSA algorithm. Using a first approximation of the scanning dynamics, the delivery times for the scan sequences before and after path optimization were calculated for comparison. In these clinical examples, the FSA optimization shortened the total scan path length for the 3D target volumes by approximately 13%-56%. The number of extraneous spilled particles was correspondingly reduced by about 13%-54% due to the more efficient scanning maps that eliminated multiple crossings through regions of zero fluence. The relative decrease in delivery time due to path length minimization was estimated to be less than 1%, due to both a high scanning speed and time requirements that could not be altered by optimization (e.g., time required to change the beam energy). In a preliminary consideration of application to rescanning techniques, the decrease in delivery time was estimated to be 4%-20%.

Original languageEnglish
Pages (from-to)3457-3464
Number of pages8
JournalMedical Physics
Volume34
Issue number9
DOIs
StatePublished - 2007
Externally publishedYes

Keywords

  • Optimization
  • Proton therapy
  • Scanning

Fingerprint

Dive into the research topics of 'Demonstration of scan path optimization in proton therapy'. Together they form a unique fingerprint.

Cite this