Quantitative analysis of morphological techniques for automatic classification of micro-calcifications in digitized mammograms

作者：

Highlights：

• This study quantitatively justified the use of a specific processing algorithm.

• Contrast operator and extended maxima thresholding produced a sensitivity of 0.9774.

• Signal Efficiency was random when varying the number of features.

• ROC area for Gaussian kernel with σ = 100 in SVM, considering 60 features, was 0.976.

摘要

•This study quantitatively justified the use of a specific processing algorithm.•Contrast operator and extended maxima thresholding produced a sensitivity of 0.9774.•Signal Efficiency was random when varying the number of features.•ROC area for Gaussian kernel with σ = 100 in SVM, considering 60 features, was 0.976.

论文关键词：Mammogram analysis,Morphological reconstruction,Digital mammography,Micro-calcification detection,Mathematical Morphology

论文评审过程：Available online 10 June 2014.

论文官网地址：https://doi.org/10.1016/j.eswa.2014.05.051