Multi-criterion mammographic risk analysis supported with multi-label fuzzy-rough feature selection

作者：

Highlights：

• By taking a novel view of the utilization of four criterions: BI-RADS, Boyd, Tabár, Wolfe, the mammographic risk assessment is conducted as a multi-label learning process.

• An associated multi-label fuzzy-rough feature selection algorithm is proposed to reduce the dimension of the data sets.

• By the nature of the four criterions, the practical significance of the association rules produced in the proposed approach is verified.

• Compared to the alternative single-label feature selection methods, the proposed algorithm provides superior performances in general.

摘要

•By taking a novel view of the utilization of four criterions: BI-RADS, Boyd, Tabár, Wolfe, the mammographic risk assessment is conducted as a multi-label learning process.•An associated multi-label fuzzy-rough feature selection algorithm is proposed to reduce the dimension of the data sets.•By the nature of the four criterions, the practical significance of the association rules produced in the proposed approach is verified.•Compared to the alternative single-label feature selection methods, the proposed algorithm provides superior performances in general.

论文关键词：Learning classifiers,Feature selection,Multiple criteria,Multiple labels,Fuzzy-rough dependency,Mammographic risk

论文评审过程：Received 10 April 2019, Revised 15 August 2019, Accepted 6 September 2019, Available online 25 September 2019, Version of Record 25 September 2019.

论文官网地址：https://doi.org/10.1016/j.artmed.2019.101722