Computer-aided diagnosis of mammographic masses based on a supervised content-based image retrieval approach.

•A two-stage CADx scheme coupling a margin-specific supervised CBIR with a decision stage.•A novel variation of HOG targeting an improved texture-based representation of the mass margin.•Use of an ensemble of SVMs aiming to transform the feature vector to a new representation vector.•Creation of an enriched DDSM subset containing accurate ground truth contour delineations.•The proposed scheme outperforms state-of-the-art CBIR-CADx schemes.