Close  
Figure 3: The receiver operating characteristic (ROC) curves for the Tissue Adequacy mode considering (a) only the optimal classification as the true positive, and (b) both the optimal and suboptimal classifications combined as the true positive. The curves suggest that the model can be reliably used for QC on the basis of tissue presence. The two yaxes provided for each receiver operating characteristic plot denote the conventional true positive rate on the left (in blue) and the corresponding threshold values on the right (in red). The latter is provided to allow the graphic determination of what AI model scores [from Figure 2] will correspond to different proportions of true and false positives 
