Share Email Print

Proceedings Paper

Effects of sample size on classifier design for computer-aided diagnosis
Author(s): Heang-Ping Chan; Berkman Sahiner; Robert F. Wagner; Nicholas Petrick
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

One of the important issues in the development of computer- aided diagnosis (CAD) algorithms is the design of classifiers. A classifier is designed with case samples drawn from the patient population. Generally, the sample size available for classifier design is limited, which introduces bias and variance into the performance of the trained classifier. Fukunaga showed that the bias on the probability of misclassification is proportional to 1/Nt, where Nt is the design (training) sample size, under conditions that the higher-order terms can be neglected. For CAD applications, a commonly used performance index for a classifier is the area, Az, under the receiver operating characteristic curve. We have studied the dependence of the bias in Az on sample size by computer simulation for a linear classifier and nonlinear classifiers such as the quadratic and the backpropagation neural network classifiers.

Paper Details

Date Published: 24 June 1998
PDF: 14 pages
Proc. SPIE 3338, Medical Imaging 1998: Image Processing, (24 June 1998); doi: 10.1117/12.310895
Show Author Affiliations
Heang-Ping Chan, Univ. of Michigan (United States)
Berkman Sahiner, Univ. of Michigan (United States)
Robert F. Wagner, FDA Ctr. for Devices and Radiological Health (United States)
Nicholas Petrick, Univ. of Michigan (United States)

Published in SPIE Proceedings Vol. 3338:
Medical Imaging 1998: Image Processing
Kenneth M. Hanson, Editor(s)

© SPIE. Terms of Use
Back to Top