Table 5 Average performance of MedMNIST2D in metrics of average AUC and average ACC over all 2D datasets.

From: MedMNIST v2 - A large-scale lightweight benchmark for 2D and 3D biomedical image classification

Methods

AVG AUC

AVG ACC

ResNet-18 (28)10

0.922

0.819

ResNet-18 (224)10

0.925

0.821

ResNet-50 (28)10

0.920

0.816

ResNet-50 (224)10

0.923

0.821

auto-sklearn11

0.878

0.722

AutoKeras12

0.917

0.813

Google AutoML Vision

0.927

0.809

Back to article page