Figure 2 | Scientific Reports

Figure 2

From: Impact of image compression on deep learning-based mammogram classification

Figure 2

Classification performance of models tested on the compressed images. Subfigure (a) is the receiver operating characteristic curve (ROC), while subfigure (b) is the precision-recall (PR) curve. Each label in the legend annotates a model trained and tested on images with a single compression ratio, and the performance for area under ROC or PR curves (AUROC or AUPRC) with 95% confidence interval in parentheses.

Back to article page