Application of Artificial Neural Networks in Breast Cancer Classification: A Comparati e Study

Nwaneri SC(1), Nwoye EO(2), Irurhe NK(3), Babatunde AM(4),


(1) Biomedical Engineering Department, College of Medicine, University of Lagos, Nigeria
(2) Biomedical Engineering Department, College of Medicine, University of Lagos, Nigeria
(3) Radiology Department, College of Medicine, University of Lagos, Nigeria
(4) Radiology Department, College of Medicine, University of Lagos, Nigeria
Corresponding Author

Abstract


Background: Breast cancer is a leading cause of death especially among women globally. The classification task of breast lump as benign or malignant is due to the experience and skill of the radiologist. However, Artificial Neural Networks (ANNs) can be developed to assist radiologists in decision-making.

Objective: The purpose of this study is to develop ANN-based models for breast cancer classification. Method: The five features of retrospective breast ultrasound data obtained from Lagos University Teaching Hospital (LUTH) consisting of 83 samples were rated using Breast Imaging Reporting and Data system (BIRADS).The data was normalized and trained in MATLAB software version (R2009a)using a feedforward multilayer ANN with 5 inputs neurons, 10 hidden neurons and one output neuron. The hidden neurons were increased in steps of 10 for different iterations to a maximum of 100 neurons in the hidden layer. The well-known Wisconsin Breast Cancer Data (WBCD) comprising 699 samples of digitized data was also trained with the same algorithm and parameters.

Results: The results show thatANNs performance in both cases was quite high. It was also proved that there was no direct relationship between the performance of the network and the number of hidden neurons.

Conclusion: ANNs are efficient classifiers that can be utilized in the diagnosis of breast cancer in the country


Keywords


Breast Cancer, Artificial Neural Networks, Breast Imaging Reporting and Data System (BI-RADS), Ultrasound, Radiologists

Full Text: PDF

Article Metrics

Abstract View : 110 times
PDF Download : 12 times

Refbacks

  • There are currently no refbacks.