Classifier ensemble for mammography CAD system combining feature selection with ensemble learning

The classifier plays an important role in the system of detecting abnormal shadows in mammograms. Normal and abnormal shadows sometimes look very similar and their differences are very subtle and complex. It is difficult to design classifier with high detection accuracy and generalization ability. I...

Full description

Saved in:
Bibliographic Details
Published in:International Congress series 2005-05, Vol.1281, p.1047-1051
Main Authors: Nemoto, Mitsutaka, Shimizu, Akinobu, Kobatake, Hidefumi, Takeo, Hideya, Nawano, Shigeru
Format: Article
Language:English
Subjects:
Abnormal shadows
Classifier
Ensemble learning
Feature selection
Mammography
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The classifier plays an important role in the system of detecting abnormal shadows in mammograms. Normal and abnormal shadows sometimes look very similar and their differences are very subtle and complex. It is difficult to design classifier with high detection accuracy and generalization ability. In this paper, we proposed the designing method using ensemble learning method (e.g. AdaBoosting method) with feature selection to solve the trade-off problem between accuracy and generalization. The result of large-scale experiments using 1793 mammograms showed that the introduction of classifier ensemble is effective in improving the performance of the mammography CAD system. Effectiveness of introducing feature selection process for each component classifier has been also shown. Those results have been confirmed by the statistical tests.
ISSN:0531-5131
1873-6157
DOI:10.1016/j.ics.2005.03.083