Enhanced K-Nearest Neighbour Firefly Model for Breast Cancer Classification System

Authors: Olusi, T.R., Balogun, M.O., Adepoju, T.M., Sotonwa, K.A., Adetunji, B.A., Olabiyisi, O.S. And Omidiora, O.E.

DOI Info: http://doi.org/10.5281/zenodo.21047523

ABSTRACT

Most of the existing classification algorithms for breast cancer are computationally intensive, have memory limitations, and are sensitive to the local structure of data. In this research, there was a development of an enhanced K-Nearest Neighbour (KNN)-based Firefly model for the categorisation of breast cancer. Breast images were acquired from the mammogram images database of the University of Ilorin Teaching Hospital. KNN and K-Nearest Neighbour-Eagle Strategy Firefly Algorithm (KNN-ESFA), which formed the enhanced KNN, were used to classify the images. The model was implemented in the Python programming language version 6.0. Evaluation of KNN and enhanced KNN was carried out using sensitivity, specificity, accuracy, Prediction Time (PT), Memory Usage (MU), and Receiver Operating Characteristics (ROC) as performance metrics. The enhanced KNN and KNN techniques were validated by carrying out a t-test to compare the difference between the two techniques at a 5% significance level. The sensitivity, specificity, accuracy, PT, MU, and ROC for KNN gave 67.00%, 74.00%, 67.92%, 309.22s, 10KB, and 0.92, respectively, while the corresponding values for enhanced KNN were 95.00%, 98.00%, 96.50%, 5.92s, 5.31KB, and 0.98, respectively. The t-test indicated that there was a wide range of differences (p=0.000) between KNN and KNN-ESFA for PT. The developed model improved breast cancer classification compared to existing methods. Therefore, the enhanced K-Nearest Neighbour-based Firefly classification model can be used in the medical field for accurate and timely breast cancer detection and classification.

Download Full Text