Skip to main content
. 2017 Jun 12;15:613–624. doi: 10.1016/j.nicl.2017.06.012

Table 4.

Comparison of the results using different classifiers and features.

Features Method Classifier Overall accuracy Balanced accuracy
Volume features Koikkalainen et al. (2016) DSI 50.4 50.7
Proposed RUSBoost 58.6 58.7
Grading features Koikkalainen et al. (2016) DSI 58.3 51.5
Proposed RUSBoost 66.6 62.7