Paper Info
Title
Mammogram Segmentation Using Rough k-Means and Mass Lesion Classification with Artificial Neural Network.
Abstract
The mammography is the most effective procedure for an early diagnosis of the breast cancer. Mammographic screening has been shown to be effective in reducing mortality rates by 30%-70%. In the analysis of Mammography images using Computer-Aided Diagnosis, Segmentation stage is one of the most Significant step, since it affects the accuracy of the Feature Extraction & Classification. In this paper, Rough k-means approach is used for segmentation of tumor from breast parenchyma. Pixel objects which definitely belong to the tumor region are classified under lower approximation, where as objects which possibly belong to the same are categorized as upper approximation. The difference of upper and lower approximation will result with objects in the rough boundaries. The segmentation algorithm has been verified on Mammo-grams from Mias database and the CICRI database. (Central India Cancer Research Institute, Nagpur, India). Geometrical and Textural features were calculated for segmented region. Once the features were computed for each region of interest (ROT), they are used as inputs to Artificial Neural Network (ANN) for classification as Benign or Malignant. Results of Rough k-means segmentation were compared with Otsu method of segmentation using ANN. Results indicate that Rough k-means method performs better than Otsu method in terms of classification accuracy up to 95% and can also reduces the number of biopsies required in the diagnostic process.
Year
DOI
Venue
2012
10.1007/978-3-642-35326-0_7
ADVANCED MACHINE LEARNING TECHNOLOGIES AND APPLICATIONS
Keywords
Field
DocType
Breast Cancer,Mammography,Rough k-means,Feature selection,Artificial Neural Network
Mammography,k-means clustering,Computer vision,Feature selection,Pattern recognition,Segmentation,Computer science,Feature extraction,Otsu's method,Artificial intelligence,Region of interest,Artificial neural network
Conference
Volume
ISSN
Citations 
322
1865-0929
0
PageRank 
References 
Authors
0.34
8
3
Name
Order
Citations
PageRank
Vibha Bafna Bora121.04
A. G. Kothari2289.96
Avinash G. Keskar3127.41