PhD Final Seminar – Cathy Hargrave

The development of a clinical decision-making framework for image guided radiation therapy

Student: Cathy Hargrave

When: Wednesday, 26th April 2017 2:00 PM-4:00 PM

Where: GP-S Block, Level 3, Room 301

Supervisors:

  • Kerrie Mengersen (Principal)
  • Fiona Harden
  • Michael Poulsen

Panel Members:

  • Kerrie Mengersen (Chair)
  • Fiona Harden
  • Michael Poulsen
  • Tomasz Bednarz
  • Nicole White

Abstract:

Radiation therapy is fundamental to the treatment of cancer. A geographic miss guarantees treatment failure, so it is critical to ensure that the target volume is accurately targeted by the treatment fields each day. Image-guided radiation therapy (IGRT) facilitates imaging the target volume using on-board imaging technology to assess the accuracy of treatment delivery. Clinical time-constraints impose a narrow window of time for IGRT image analysis to be performed. However, IGRT decision-making has become more complex with the introduction of 3D on-board imaging. This thesis aims to develop a framework to inform and support efficient clinical IGRT decision-making, particularly online IGRT where 3D imaging is utilised. Bayesian networks (BN) are ideally suited for modelling complex systems, with outputs that can predict or provide advice on outcomes of interest.

A conceptual model for the IGRT decision-making BN, able to be adapted to different IGRT technologies and techniques, is developed using the literature and elicited expert opinion. To demonstrate how the BN can be adapted for specific cancer sites, technologies and radiotherapy treatment techniques, it is developed and quantified for online cone beam computed tomography (CBCT)-based IGRT for prostate cancer. An image feature alignment score (FAS) for online IGRT decision-making is developed using classification and regression tree (CART) and boosted regression tree (BRT) analysis of a retrospective dosimetric evaluation of delivered treatments and the results of a prostate IGRT practice survey. Volume overlap and surface distance metrics predictive of treatment plan compliance are included in weighted functions used to calculate the FAS. Conditional probability tables for the BN are derived using data from modelling of the FAS, the survey and reported studies. Mollweide projections, used to calculate daily tumour and organs at risk directional differences, are used in conjunction with the outputs of the BN to complete the clinical IGRT decision-making framework.

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s