Open Access Open Access  Restricted Access Subscription Access
Open Access Open Access Open Access  Restricted Access Restricted Access Subscription Access

Discrete Shearlet Transform based Mass Classification System for Digital Mammograms


Affiliations
1 St.Peter’s University, Chennai, Tamil Nadu, India
2 Dr.M.G.R.University (phase II), Chennai, Tamil Nadu, India
     

   Subscribe/Renew Journal


In this paper, Discrete Shearlet Transform (DST) basedmass classification system for digital mammogram is developed. Therecent enhancement in multi resolution analysis is DST whichdiminishes the disadvantage of the wavelets that they are not veryeffective if the images containing distributed discontinuities such as edges. Initially, the given mammogram is decomposed by using DSTwith various directions. The features used in depict the masses are theenergies of all directional sub-bands of the decomposed image. In theproposed method 2-level decomposition with 2 to 64 directions areused to extract the features. In the classification stage, Support VectorMachine (SVM) classifier with two levels is proposed. In the first onethe given unknown mammogram is classified into normal or abnormalcategory and finally the abnormal severity is classified into benign ormalignant. Experiments are conducted on Mammography ImageAnalysis society (MIAS) database. The average classificationaccuracy achieved for normal/abnormal is 88.72% andbenign/malignant is 94.74%.


Keywords

Shearlet Transform SVM Classifier, Digital Mammograms, Mass, Benign and Malignant.
User
Subscription Login to verify subscription
Notifications
Font Size

Abstract Views: 228

PDF Views: 3




  • Discrete Shearlet Transform based Mass Classification System for Digital Mammograms

Abstract Views: 228  |  PDF Views: 3

Authors

J. Amjath Ali
St.Peter’s University, Chennai, Tamil Nadu, India
J. Janet
Dr.M.G.R.University (phase II), Chennai, Tamil Nadu, India

Abstract


In this paper, Discrete Shearlet Transform (DST) basedmass classification system for digital mammogram is developed. Therecent enhancement in multi resolution analysis is DST whichdiminishes the disadvantage of the wavelets that they are not veryeffective if the images containing distributed discontinuities such as edges. Initially, the given mammogram is decomposed by using DSTwith various directions. The features used in depict the masses are theenergies of all directional sub-bands of the decomposed image. In theproposed method 2-level decomposition with 2 to 64 directions areused to extract the features. In the classification stage, Support VectorMachine (SVM) classifier with two levels is proposed. In the first onethe given unknown mammogram is classified into normal or abnormalcategory and finally the abnormal severity is classified into benign ormalignant. Experiments are conducted on Mammography ImageAnalysis society (MIAS) database. The average classificationaccuracy achieved for normal/abnormal is 88.72% andbenign/malignant is 94.74%.


Keywords


Shearlet Transform SVM Classifier, Digital Mammograms, Mass, Benign and Malignant.