Encrypted login | home

Program Information

ROC: A New Method in Radiotherapy QA


E Krupinski

E Boehnke
no image available
L Archambault




E Krupinski1*, E Boehnke2*, L Archambault3*, (1) Emory University, Atlanta, GA, (2) Cedars-Sinai Medical Center, Los Angeles, CA, (3) CHUQ Pavillon Hotel-Dieu de Quebec, Quebec, QC

Presentations

4:30 PM : ROC Methods in Diagnostic Imaging - E Krupinski, Presenting Author
5:00 PM : Characterization of a novel detector using ROC - E Boehnke, Presenting Author
5:30 PM : A framework for utilising ROC methodology in radiotherapy QA - L Archambault, Presenting Author

WE-G-FS4-0 (Wednesday, August 2, 2017) 4:30 PM - 6:00 PM Room: Four Seasons 4


In this session, we will introduce the methods of decision theory in patient specific quality assurance. Receiver Operating Characteristic (ROC) techniques have been used in diagnostic imaging since shortly after WWII and have undergone a variety of advances to address the different types of tasks and issues involved in clinical image decision making. The current challenges in this area and the types of ROC analysis that have been developed to address them, including LROC, FROC, AFROC and JAFROC will be discussed.

In the patient specific QA domain, ROC curves can be used to synthesize an unbiased decision-making framework. By distilling a device’s response into sensitivity and specificity, we are able to compare systems which are measuring the same thing (treatment fluence), but giving different types of outputs, e.g. a point-based measurement vs a gamma analysis. The flexibility of ROC analysis applied to QA is demonstrated with a new large area ion chamber. By framing the output in terms of sensitivity and specificity to known errors, new benchmarks can be made which not only characterize the abilities of the device, but also provide guidance for clinical use. The flexibility of ROC analysis allows it to be used across a broad spectrum of QA techniques, acting as a powerful tool to evaluate the innovations of today and the future.

After reviewing recent applications of ROC-analysis in radiation therapy, we will propose a general methodology for applying the ROC framework to quality assurance (QA). We will then illustrate how this unbiased approach can generally improve the QA process by improving criterion optimization and helping choose the right detector for the right task. We will finally open the discussion on the future role of ROC-analysis in the medical physicist toolbox.

Learning Objectives:
1. To understand how imaging methodology can be applied in the patient specific QA domain.
2. To understand clinical image decision making tools and review current challenges in ROC analysis.
3. To understand novel uses of ROC methods in quantifying the response of a novel detector.
4. To explain a general methodology for applying the ROC framework to quality assurance (QA).

Handouts


Contact Email: