|
Activity Number:
|
259
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 13, 2002 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section*
|
| Abstract - #301642 |
|
Title:
|
Comparing Medical Estimation Methods without Truth
|
|
Author(s):
|
John Hoppin*+ and Matthew Kupinski and Eric Clarkson and Harrison Barrett
|
|
Affiliation(s):
|
University of Arizona and University of Arizona and University of Arizona and University of Arizona
|
|
Address:
|
PO Box 245067, Tucson, Arizona, 85724-5067, USA
|
|
Keywords:
|
Estimation ; Gold-standard ; Maximum likelihood ; Regression
|
|
Abstract:
|
Often in medical imaging we estimate a parameter of interest for a patient to assist with diagnosis. Many different estimation methods may exist; rarely can one be considered a gold standard. Due to this lack of a gold standard, it is difficult to evaluate and compare different estimation methods. We present a method of evaluating different estimation methods without the use of a gold standard. This method is equivalent to fitting regression lines without the x-axis. To use this method we must have multiple estimates of the clinical parameter of interest for each patient of a given population. We assume the statistical distribution for the true values of the clinical parameter of interest is a member of a given family of parameterized distributions. Using the observed data, we estimate the model parameters and the parameters characterizing the distribution of the clinical parameter. These estimates can be used to rank the systems for the given estimation task. We will compare and discuss the relationship between this technique and previous techniques, such as Bland-Altman plots and latent variable analysis.
|
