Optimization of computer aided detection systems: An evolutionary approach

作者：

Highlights：

• Genetic algorithms selected optimal parameters for breast masses detection.

• A general-purpose multi-objective framework balanced sensitivity and specificity.

• We defined a modified asymmetric Dice coefficient to avoid over-segmentation.

• Association rule mining characterized the effect and significance of each parameter.

摘要

•Genetic algorithms selected optimal parameters for breast masses detection.•A general-purpose multi-objective framework balanced sensitivity and specificity.•We defined a modified asymmetric Dice coefficient to avoid over-segmentation.•Association rule mining characterized the effect and significance of each parameter.

论文关键词：Computer aided detection,Genetic algorithms,Breast tomosynthesis,Segmentation,Optimization

论文评审过程：Received 11 August 2017, Revised 23 December 2017, Accepted 19 January 2018, Available online 31 January 2018, Version of Record 16 February 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.01.028