Activity Number:
|
22
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, August 11, 2002 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Biometrics Section*
|
Abstract - #300894 |
Title:
|
Classifying Populations from Samples Using Quantitatve Pathology
|
Author(s):
|
Richard Swartz*+ and Loyd West and Iouri Boiko and Anais Malpica and Martial Guillaud and Michele Follen and Dennis Cox
|
Affiliation(s):
|
Rice University and Naval Medical Center and U. T. M. D. Anderson Cancer Center and U. T. M. D. Anderson Cancer Center and British Columbia Cancer Research Center and U. T. M. D. Anderson Cancer Center and Rice University
|
Address:
|
, Houston, Texas, ,
|
Keywords:
|
Logistic Regression ; Quantitative Pathology
|
Abstract:
|
This study develops a method that discriminates between normal and cancerous tissue slices (i.e., populations of cells) using statistical methods applied to quantitative measurements made on a sample of cells. Quantitative image analysis procedures produce high-dimensional feature vectors taken from cell images. We first use logistic regression to summarize the cellular features with a probability score and then fit a beta density to the probabilities associated with a given sample, using maximum likelihood estimation to create population features. In addition, we created a biologically motivated population feature using the DNA Index. These population features are then used to develop different scores for classification of the population, which were compared using ROC curve analysis. The method was tested using data from cervical adenocarcinomas. The score which maximizes the sum of the specificity and sensitivity achieved a sensitivity of 96.4% with a specificity of 92.3%.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |