Automated Detection of Breast Cancer Using Artificial Neural Networks and Fuzzy Logic

  • Esraa A. AL-Dreabi Department of Computer Science, King Abdullah II School for Information Technology, University of Jordan, Amman, Jordan
  • Mohammad M. Otoom Department of Computer Science, College of Science in Zulfi, Majmaah University, Riyadh, Saudia Arabia
  • Bareqa Salah Division of plastic and reconstructive surgery, Faculty of Medicine, University of Jordan, Amman, Jordan
  • Ziad M. Hawamdeh Department of Rehabilitation Medicine, Faculty of Medicine, University of Jordan, Amman, Jordan
  • Mohammad Alshraideh Department of Computer Science, King Abdullah II School for Information Technology, University of Jordan, Amman, Jordan
Keywords: Benign, Breast cancer, Fuzzy system, Malignant, Neural Networks.

Abstract

Our aim was to develop a diagnostic system that could classify breast tumors as either malignant or benign to provide a faster and more reliable method for patients. In order to accomplish this, we built two systems: one is based on Artificial Neural Networks (ANN) with a resilient back propagation and the other is based on fuzzy logic. We used the dataset provided by the University of California Irvine (UCI) Machine Learning Repository: the Wisconsin Diagnostic Breast Cancer (WDBC) dataset which describes characteristics of the cell nuclei presented in the images. The dataset is composed of features computed from digitized images of a Fine Needle Aspirate (FNA) of the breast mass. The system is based on ANN and was built using a feed-forward neural network with a Resilient Back Propagation (Rprop) algorithm that used to train the network, the number of hidden layers and hidden neurons determined by performing experiments and selecting the highest architectural accuracy. In order to obtain general architecture and to identify the accuracy of this system, we used ten-folds cross validation. The second system is based on fuzzy logic, and we built a Fuzzy Inference System (FIS). The decision tree was used to define the membership functions and the rules. The experiments were performed on two types of FIS: Sugeno-type and Mamdani-type. For the system based on ANN, Feed-Forward Neural Network presented the highest accuracy at 97.6%. While for fuzzy system, Sugeno FIS showed the highest accuracy at 94.8%. Since breast tumors, both malignant and benign, share structural similarities, the process of their detection is extremely difficult and time consuming if it is to be manually classified. Laboratory analysis or biopsies of the tumor is a manual, time consuming process yet it is accurate system of prediction. It is, however, prone to human errors. Consequently, a need of creating an automated system to provide a faster and more reliable method of diagnosis and prediction for patients is rising. In this paper, we developed two kinds of artificial intelligence systems that can help physicians to classify breast cancer tumors as either malignant or benign.

References

Kononenko I. Machine learning for medical diagnosis: history, state of the art and perspective, Artificial Intelligence in Medicine. Slovenia: Ljubljana; 2001. p. 89-109.

American Cancer Society, Breast Cancer Overview. USA: 2010.p. 1-58.

Frank A, and Asuncion A. UCI Machine Learning Repository, Irvine, CA: University of California, School of Information and Computer Science http://archive.ics.uci.edu/ml/datasets/Breast+Cancer+Wisconsin+%28Diagnostic%29, Accessed: July 16th, 2011

Sewak M, Vaidya P, Chung Chan C, and Duan Z. SVM Approach to Breast Cancer Classification, IMSCCS '07 Proceedings of the Second International Multi-Symposiums on Computer and Computational Sciences. 2007; 94.

Yan H, Jiang Y, Zheng J, Peng C, and Li Q, A Multilayer Perceptron-Based Medical Decision Support System for Heart Disease Diagnosis, Expert Systems with Applications. 2006; 30: 272-281.

Xu S, and Chen L. A Novel Approach for Determining the Optimal Number of Hidden Layer Neurons for FNN?s and Its Application in Data Mining, In the 5th International Conference on Information Technology and Applications, Cairns, Australia. 2008; 683-686.

Hassanien A. Classification and feature selection of breast cancer Data based on Decision Tree Algorithm, International Journal of Studies in Informatics and Control Journal. 2003;12:33-39.

Korting T. C4.5 algorithm and Multivariate Decision Trees, Multivariate decision trees, Machine Learning, 2006; 19:45

Demuth H, Beale M, and Hagan M, Math Works, Neural Network Toolbox

Liu L, and Deng M. An Evolutionary Artificial Neural Network Approach for Breast Cancer Diagnosis, Third International Conference on Knowledge Discovery and Data Mining, 2010.

Anagnostopoulos I, and Maglogiannis I. Neural network-based diagnostic and prognostic estimations in breast cancer microscopic instances, International Federation for Medical and Biological Engineering. 2006; 44:773

Mashor M, Esugasini S, Isa N and Othman N. Classification of Breast Lesions Using Artificial Neural Network, Biomed 06, IFMBE Proceedings. 2007; 15: 45-49.

Parthiban L, and Subramanian R, CANFIS

Balanica V, Dumitrache L, Caramihai M, Rae W, and Herbst C. Evaluation Of Breast Cancer Risk By Using Fuzzy Logic, U.P.B. Sci. Bull, Series C. 2011;73: 402-4014.

Published
2017-08-26
Section
Articles