European Radiology Experimental

Table 3 Results of the predictions of the deep learning (DL) model and readers using the labels derived from digital subtraction angiography as the ground truth

From: Detection of femoropopliteal arterial steno-occlusion at MR angiography: initial experience with artificial intelligence

Reader	Side	Accuracy	ROC-AUC
DL model	Left	0.897	0.913
DL model	Right	0.743	0.830
Reader 1	Left	0.931	0.930
Reader 1	Right	0.800	0.799
Reader 2	Left	0.862	0.861
Reader 2	Right	0.771	0.771

ROC-AUC Area under the curve at receiver operating characteristics analysis

Back to article page