Proficient feature extraction strategy for performance enhancement of NN based early breast tumor detection

Abstract

Ultra Wideband is one of the promising microwave imaging techniques for breast tumor prognosis. The basic principle of tumor detection depends on the dielectric properties discrepancies between healthy and tumorous tissue. Usually, the tumor affected tissues scatter more signal than the healthy one and are used for early tumor detection through received pulses. Feedforward backpropagation neural network(NN) was so far used for some research works by showing its detection efficiency up to 1mm (radius) size with 95.8% accuracy. This paper introduces an efficient feature extraction method to further improve the performance by considering four main features of backpropagation NN. This performance is being increased to 99.99%. This strategy is well justified for classifying the normal and tumor affected breast with 100% accuracy in its early stage. It also enhances the training and testing performances by reducing the required duration. The overall performance is 99.99% verified by using thirteen different tumor sizes.