Tumor motion prediction with the diaphragm as a surrogate: A feasibility study

Laura I. Cerviño, Yan Jiang, Ajay Sandhu, Steve B. Jiang

Research output: Contribution to journalArticle

18 Scopus citations

Abstract

We have previously assessed the use of the diaphragm as a surrogate for predicting real-time tumor position with linear models built with training data extracted from the same treatment fraction (Cervĩo et al 2009 Phys. Med. Biol. 54 3529-41). However, practical use in the clinical setting requires the capability of predicting tumor position throughout the treatment course using a model built at the beginning of the course. We evaluate the inter-fraction applicability of linear models to predict superior-inferior tumor position based on diaphragm position using 21 fluoroscopic sequences from five lung cancer patients. Tumor position is predicted with models built during the first fluoroscopic sequence of each patient. Other fluoroscopic sets are registered to the first set with five different methods. The mean localization prediction error and maximum error at a 95% confidence level averaged over all patients are found to be 1.2 mm and 2.9 mm, respectively, for bony registration and 1.2 mm and 2.8 mm, respectively, for registration based on the mean position of the tumor in the first two breathing cycles. Other registration methods produce larger prediction errors. In the clinical setting, this prediction error could be added as a margin to the target volume. We therefore conclude that it is feasible to predict lung tumor motion with diaphragm with sufficient accuracy in the clinical setting.

Original languageEnglish (US)
Pages (from-to)N221-N229
JournalPhysics in medicine and biology
Volume55
Issue number9
DOIs
StatePublished - Apr 27 2010

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging

Fingerprint Dive into the research topics of 'Tumor motion prediction with the diaphragm as a surrogate: A feasibility study'. Together they form a unique fingerprint.

  • Cite this