On the use of contrastive learning for standard-plane classification in fetal ultrasound imaging

Giovanna Migliorelli, Maria Chiara Fiorentino, Mariachiara Di Cosmo, Francesca Pia Villani, Adriano Mancini, Sara Moccia

Abstract: Highlights•Explore contrastive learning for reducing the need of large annotated fetal US datasets.•Outline contrastive learning effect for low and high inter-class variability data.•Contrastive learning benefits from transfer learning from natural image.•Contrastive learning improves classification of small datasets with low inter-class variability.