BME Frontiers / 2022 / Article / Tab 2

Research Article

A Low-Cost High-Performance Data Augmentation for Deep Learning-Based Skin Lesion Classification

Table 2

The performance of models on ISIC 2018 challenge legacy leaderboard (rows 1–3), live leaderboard (rows 4–11), and our proposed approach.

Team/authorsUse external dataUse ensemble modelsBACCSensitivity for melanomaAvg. AUCAvg. specificity

Nozdryn et al. [18]YesYes0.8850.7600.9830.833
Gessert et al. [26]YesYes0.8560.8010.9870.984
Zhuang et al.NoYes0.8450.7020.9780.980
MinjieYesYes0.8950.7780.9820.981
Amirreza et al.YesYes0.8740.5850.9790.992
IPM_HPC.YesYes0.8660.8300.9760.976
Amirreza et al. [27]YesYes0.8620.981
Mahbod et al.NoNo0.8360.7190.9750.982
ND Reddy et al.NoNo0.7350.5440.9450.968
Our approachNoNo0.8530.7890.9750.973