Application of neural networks and genetic algorithms in breast cancer screening

The aim of this work is to analyze breast cancer screening images. Firstly, genetic algorithms are used to segment the images and identify high-density areas within the mammograms. Afterwards, Convolutional Neural Networks (CNNs) are developed, trained, and tested for detecting cancerous lesions in mammography using heterogeneous images acquired with different imaging systems. The algorithms are also tested independently on different datasets and then applied to images with different vendor styles, which have been previously standardized using a neural style transfer method.