Density-wise two stage mammogram classification using texture exploiting descriptors
作者：
Highlights：
• We propose two new feature extraction descriptors for Mammogram images. These descriptors capture the texture of the breast for normal–abnormal and benign–malignant classifications.
• We perform density (breast) wise classifications, which has not been done.
• We test our algorithm on all images of IRMA (Image Retrieval in Medical Applications) database provided by RWTH Aachen, Germany. In the past people have tested their work only on a subset of these images.
• We achieve higher accuracy than existing techniques for all images (more than 92%).
摘要
•We propose two new feature extraction descriptors for Mammogram images. These descriptors capture the texture of the breast for normal–abnormal and benign–malignant classifications.•We perform density (breast) wise classifications, which has not been done.•We test our algorithm on all images of IRMA (Image Retrieval in Medical Applications) database provided by RWTH Aachen, Germany. In the past people have tested their work only on a subset of these images.•We achieve higher accuracy than existing techniques for all images (more than 92%).
论文关键词：Mammogram,Gabor filter,Histogram of gradients,Discrete Cosine Transform,Feature selection
论文评审过程：Received 10 May 2017, Revised 31 December 2017, Accepted 16 January 2018, Available online 31 January 2018, Version of Record 3 February 2018.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.01.024
