Transfer Learning Assisted Classification of Artefacts Removed and Contrast Improved Digital Mammograms

Mammograms are essential radiological images used to diagnose breast cancer well in advance. However, an accurate diagnosis also depends on the quality of mammogram images. Therefore, removal artefacts and enhancement necessary pre-processing steps. Artefact helps exclude unsolicited regions mammograms limits search for suspicious without excessive impact from background. Mammogram enhancements improve apparent visual details some features image. In this paper, we propose a method pre-processing. These pre-processed then fed into Deep Convolutional Neural Network classification process. Two approaches compared classify mammograms; Training model scratch Transfer Learning. Learning is excellent approach dealing with small-sized training set, allowing us consume extendibility deep learning entirely. By employing VGG16 as pre-trained network MIAS dataset, improved accuracy (96.14\%) developed other strategies described literature.