Affiliations: Department of Computer Science and Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh 221005, India
Corresponding author: Vibhav Prakash Singh, Department of Computer Science and Engineering, Indian Institute of Technology (BHU), Varanasi, Uttar Pradesh 221005, India. E-mail: [email protected]
Abstract: Early diagnosis of breast cancer can improve the survival rate by detecting cancer at initial stage. In this paper, an efficient content-based mammogram retrieval system is proposed, which helps in early diagnosis of breast cancer by classifying the current case mammogram and retrieving similar past cases mammograms already annotated by diagnostic descriptions and treatment results. The proposed steps include cropping of mammograms for finding the region of interest (ROI), feature extraction using wavelet based-complete local binary pattern (W-CLBP) and K-means clustering. Strong texture characteristics of the mammogram are captured using CLBP to all detailed coefficients (LH, HL, and HH) from two level decomposed 2 Dimensional-discrete wavelet transform (2D-DWT) of ROI. Further, K-means generates the clusters based on this texture similarity of mammograms, and query mammogram features are matched with all cluster representatives to find the closest cluster. Finally, images are retrieved from this closest cluster using Euclidean distance similarity measure. So, at the searching time the query mammogram is searched only in small sub-set depending upon the cluster size and is not compared with all the images in the database, reflects a superior response time with good retrieval performances. Experiments on benchmark mammography image analysis Society (MIAS) database confirm that the proposed method has better say with respect to other four variants of texture features.