Multivariate statistical modelling to improve particle treatment verification: Implications for prompt gamma-ray timing

Research output: Contribution to journalResearch articleContributedpeer-review

Contributors

Abstract

We present an improved method for in-vivo proton range verification by prompt gamma-ray timing based on multivariate statistical modelling. To this end, prompt gamma-ray timing distributions acquired during pencil beam irradiation of an acrylic glass phantom with air cavities of different thicknesses were analysed. Relevant distribution features were chosen using forward variable selection and the Least Absolute Shrinkage and Selection Operator (LASSO) from a feature assortment based on recommendations of the Image Biomarker Standardisation Initiative. Candidate models were defined by multivariate linear regression and evaluated based on their coefficient of determination R2 and root mean square error RMSE. The newly developed models showed a clearly improved predictive power (R2 > 0.7) compared to the previously used models (R2 < 0.5) and allowed for the identification of introduced air cavities in a scanned treatment field. These results demonstrate that elaborate statistical models can enhance prompt gamma-ray based treatment verification and increase its potential for routine clinical application.

Details

Original languageEnglish
Article number932950
JournalFrontiers in physics
Volume10
Publication statusPublished - 19 Aug 2022
Peer-reviewedYes

External IDs

ORCID /0000-0002-7017-3738/work/146646040

Keywords

Keywords

  • machine learning, multivariate modelling, prompt gamma-ray timing, proton therapy, treatment verification