|
Title:
|
MIDPDC: A NEW FRAMEWORK TO SUPPORT DIGITAL MAMMOGRAM DIAGNOSIS |
|
Author(s):
|
J.senthilkumar , A.ezhilarasi , D.manjula |
|
ISBN:
|
978-972-8924-88-1 |
|
Editors:
|
Ajith P. Abraham |
|
Year:
|
2009 |
|
Edition:
|
Single |
|
Keywords:
|
Pattern Discovery, Discretisation, Feature selection, Image diagnosis |
|
Type:
|
Short Paper |
|
First Page:
|
121 |
|
Last Page:
|
126 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
In this paper, we present a framework based on pattern clustering to help diagnosis of mammogram abnormalities. Our
framework - MIDPDC - combines visual features automatically extracted from mammogram images with high-level
knowledge obtained from specialists to mine pattern, suggesting possible diagnoses. The proposed method incorporates
two new algorithms called PreSPG and PDC. Our framework is optimized, in the sense PreSPG algorithm combines, in a
single step, feature discretisation and feature selection, and reduces the mining complexity. PDC is a new pattern
clustering algorithm which discovers the patterns from previously unknown regularities in the data and clustering of the
patterns instead of data. The proposed method uses KNN algorithm, which is a diagnosis engine that classifies the
mammogram images. The MIDPDC system was applied to real dataset and the results presented high accuracy (up to
97.72 %) and high sensitive (up to 96%), allowing us to claim that pattern clustering with efficient feature selection can
effectively aid in the diagnosing task. |
|
|
|
|
|
|
