Skip to main content
. Author manuscript; available in PMC: 2019 Aug 9.
Published in final edited form as: Med Image Anal. 2019 Apr 22;55:136–147. doi: 10.1016/j.media.2019.04.009

Fig. 9.

Fig. 9

ROC curves for the LRCN-based motion artefact detection approach using curriculum learning. Gradually introducing harder samples during training improves the performance of the algorithm compared to the random (control-curriculum) and harder-to-easier configurations (anti-curriculum).