Agreement by concordance and reader studies
View Presentation *Brandon D Gallas, FDA/CDRH/OSEL/DIAM Keywords: reader agreement, concordance, imaging This talk discusses how concordance generalizes the area under the ROC curve (AUC) when truth is multi-level or unknown. Some examples of multi-level truth are cell expression, time to an event, and truth by an expert panel. When truth is unknown, it is often useful to investigate reader agreement and variability. This talk compares concordance to other agreement measures using simulation, characterizing behavior with respect to data transformations such as rescaling, shifting (bias), and binning. Finally, this talk will describe results from a study where pathologists scored HER2/neu immunohistochemical expression with a continuous scale and a 3-level scale. The results showed that two computer extracted features reduced reader variability, readers used the continuous scale effectively, and the continuous scale increased the precision of the agreement measures.
|
Key Dates
-
April 30 - May 22, 2013
Invited Abstract Submission Open -
June 4, 2013
Online Registration Opens -
August 9 - August 23, 2013
Invited Abstract Editing -
August 23, 2013
Short Course materials due from Instructors -
August 26, 2013
Housing Deadline -
September 9, 2013
Cancellation Deadline and Registration Closes @ 11:59 pm EDT -
September 16 - September 18, 2013
Marriott Wardman Park, Washington, DC