Title
ROI Segmentation Using Two-Fold Image with Super-Resolution Technique
Publication Name
Lecture Notes on Data Engineering and Communications Technologies
Abstract
Region of Interest (ROI) segmentation is one of the challenging steps during breast cancer detection. By calculating the threshold value, a binary image is constructed and is named a concealed image, where value 1 represents the presence of texture. With the help of a concealed image, a two-fold image is constructed, and to convert this image from low resolution to high-resolution Super-resolution technique is applied. This constructed high-resolution image can be used during developing Computer Aided Diagnosis Systems for breast cancer detection. The efficiency of the proposed approach is tested on the suspicious patches of the IRMA reference dataset. The testing of the work is performed on 762 ROIs, where 352 are from the benign class, and 410 are from the malignant class. The experiments have shown that the proposed two-fold image has attained the values 0.926 and 0.956 for quality measures for benign and malignant classes, respectively. A comparative analysis of our proposed method with two existing and similar methodologies also validates the correctness and accuracy of our result.
Open Access Status
This publication is not available as open access
Volume
175
First Page
323
Last Page
334