English

Abstract

Women have been affected by many diseases for decades due to their low immune systems. Especially breast cancer is the second largest and most harmful disease for a woman that, leads to death. The earlier prediction of breast cancer can help for easy treatment and Save lives. Deep Learning methods are implemented in medical fields to attain an effective prediction and higher prediction accuracy. Therefore, in this work, the novel hybrid prediction topology is implemented two feature extractions, feature fusion, feature selection and classification with several algorithms such as Linear Discriminant Analysis (LDA), Canonical Correlation Analysis (CCA), Convolutional Neural Network (CNN) Long Short-Term Memory (LSTM), Transit Search Optimization (TSO), Support Vector Machine (SVM) respectively. The LDA and CNN-LSTM method is used to execute a two-feature extraction, and the CCA model is used to merge the two extracted features, a feature fusion task. Next, the TSO-based feature selection is implemented to minimize the redundant features. At last, the classification is carried out using an SVM method to acquire an efficient breast cancer prediction. The proposed hybrid method has achieved more accurate decision-making and efficient training than the conventional methods. The experimental result of the novel hybrid method acquired a superior performance in terms of precision, Recall, Accuracy, F1 Score, RMSE and MAE as 98.81%, 99.02%, 98.18%, 98.85%, 1.001 and 1.016, respectively.