Skip to main content
. 2022 Apr 28;35:103027. doi: 10.1016/j.nicl.2022.103027

Fig. 4.

Fig. 4

Performance (bootstrapped at 1000 random samples of the available test set) of individual models in comparison with observers using FROC curve of Sensitivity over average FPs per TBI case. 1) Comparison of all models. 2) Comparison of the classification models. 3) Comparison of classification and segmentation CNN with equivalent architecture. 4) Comparison of the segmentation models.