Fig. 1 | Scientific Reports

Fig. 1

From: Integrating snapshot ensemble learning into masked autoencoders for efficient self-supervised pretraining in medical imaging

Fig. 1

Comparison of performance metrics for MAE (X-ray) and Snap-MAE (X-ray) models across various masking ratio for multi-labeled pediatric thoracic disease classification. Each subplot presents a different metric (AUC, AUPRC, sensitivity, precision, and F1-score) with error bars, illustrating performance trends as the masking ratio varies from 65 to 90%.

Back to article page