Filtering and Statistical Feature based Automatic Detection of Breast Lesions in Ultrasound Images
Telagarapu Prabhakar1, S.Poonguzhali2
1Telagarapu Prabhakar, Department of ECE, GMR Institute of Technology, GMR Nagar, Rajam, Srikakulam, (Andhra Pradesh), India.
2S.Poonguzhali, Department of ECE, College of Engineering, Guindy, Anna University, Chennai (Tamil Nadu), India.
Manuscript received on 05 January 2019 | Revised Manuscript received on 13 January 2019 | Manuscript published on 30 January 2019 | PP: 63-67 | Volume-8 Issue-3, January 2019 | Retrieval Number: C2577018319/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The objective of this paper was proposed to develop a method for a fully automatic detection of breast tumor accurately. In this reduction of speckle noise was done by utilizing Curvelet, Shearlet. The filtered images acted as an input of segmentation in which the contour was initially recognized by statistical features and the region was segmented automatically. The performance analysis was done by comparing the output of the automatic segmentation region algorithm and the ground truth (Segment by radiologist).The Shearlet filtered images gave a high performance with the an accuracy of 91.51%, sensitivity of 92.24%, specificity of 89.44 %, Jaccard of 86.02 % and Dice Similarity of 91.27 % when compared to other filtered images and input image.
Keyword: Breast Ultrasound Image, Speckle Reduction, Statistical Features, Automatic Segmentation.
Scope of the Article: Image Security