 This paper proposes a novel computer-aided diagnostics, CAD, system for breast cancer diagnosis. It combines deep learning techniques with traditional handcrafted features to improve accuracy and reduce false positives. The system was tested on two databases of mammograms and ultrasounds, achieving an overall accuracy of 97.6%. This article was authored by Clara Cruz Ramos, Oscar Garcia Avila, Jose Augustin Almarez Damien, and others.