Binary Particle Swarm Optimization Based Feature Selection (BPSO-FS) for Improving Breast Cancer Prediction
- 6 Downloads
Breast cancer is currently one of the leading causes of cancer-related deaths among women around the world. Although the severity of the disease is undeniable, an efficient early diagnosis of the disease can lead to a much higher chance of survival for the patients. Effective clinical decision support systems (CDSS) could potentially be of very high utility for medical practitioners, in this regard. In this paper, a binary Particle Swarm Optimization (BPSO) based feature selection approach is presented, which can be used to improve the performance of automatic breast cancer prediction CDSS. The key idea is to formulate the problem of feature selection in terms of a discrete optimization problem, with appropriate data-driven objective function. The average cost of bad feature selection is considered as the objective function to minimize, in this work. Multiple evaluation metrics like average prediction accuracy, sensitivity, specificity and area under the curve (AUC) of the receiver operating characteristics (ROC) curve have been considered in this work, to evaluate the performance of the prediction system. Average prediction accuracies of 80.83% and 98.24% have been observed respectively for the Breast Cancer Coimbra Dataset (BCCD) and Breast Cancer Wisconsin Diagnostic Dataset (BCWDD) after feature selection is performed. When pre-feature selection and post feature selection performances are compared, overall improvements of 4–8% and 1–4% have been observed across all the evaluation metrics for BCCD and BCWDD respectively, suggesting the potential applicability of the proposed approach in real diagnostic settings.
KeywordsBreast cancer Feature selection Machine learning Clinical decision support system Binary particle swarm optimization
The authors would like to acknowledge TEQIP-III, NIT Silchar for their support.
- 1.WHO breast cancer statistics, World Health Organization breast cancer diagnosis and screening page (2019). https://www.who.int/cancer/prevention/diagnosis-screening/breast-cancer/en/. Last accessed 21 Oct 2019
- 2.U.T.O.E.D., of group, BC, First results on mortality reduction in the UK trial of early detection of breast cancer. Lancet 332(8608), 411–416 (1988)Google Scholar
- 9.J. Wang, X. Yang, H. Cai, W. Tan, C. Jin, L. Li, Discrimination of breast cancer with microcalcifications on mammography by deep learning. Sci. Rep. 6, 27327 (2016). https://doi.org/10.1038/srep27327
- 14.R. Ramos-Polln, M.A. Guevara-Lpez, C. Surez-Ortega, G. Daz-Herrero, J.M. Franco-Valiente, M. Rubio-Del-Solar, I. Ramos, Discovering mammography-based machine learning classifiers for breast cancer diagnosis. J. Med. Syst. 36(4), 2259–2269 (2012). https://doi.org/10.1007/s10916-011-9693-2CrossRefGoogle Scholar
- 15.P. Jiang, J. Peng, G. Zhang, E. Cheng, V. Megalooikonomou, H. Ling, Learning-based automatic breast tumor detection and segmentation in ultrasound images, in ISBI, pp. 1587–1590 (2012, May). https://doi.org/10.1109/isbi.2012.6235878
- 19.A. Addeh, H. Demirel, P. Zarbakhsh, Early detection of breast cancer using optimized ANFIS and features selection, in 2017 9th International Conference on Computational Intelligence and Communication Networks (CICN), IEEE, pp. 39–42 (2017, Sept). https://doi.org/10.1109/cicn.2017.8319352
- 20.C. Baneriee, S. Paul, M. Ghoshal, A comparative study of different ensemble learning techniques using wisconsin breast cancer data set, in 2017 International Conference on Computer, Electrical Communication Engineering (ICCECE), IEEE, pp. 1–6 (2017, Dec). https://doi.org/10.1109/iccece.2017.8526215
- 25.J. Kennedy, R.C. Eberhart, A discrete binary version of the particle swarm algorithm, in 1997 IEEE International Conference on Systems, Man, and Cybernetics. Computational Cybernetics and Simulation, vol. 5 (IEEE, pp. 4104–4108) (1997, Oct)Google Scholar