Transparency of deep neural networks for medical image analysis: A review of interpretability methods | Synapse