Table 2 Segmentation performance of U-Net based models without/with data-pairing.

From: Automated volumetric analysis of the inner ear fluid space from hydrops magnetic resonance imaging using 3D neural networks

Methods

MRC

Mi2

DSC

IoU

DSC

IoU

2D U-Net

0.8844

0.7936

0.6227

0.5566

3D U-Net

0.9100

0.8670

0.7634

0.6475

2D nnU-Net

0.9505

0.9075

0.8189

0.6974

3D nnU-Net

0.9560

0.9175

0.8327

0.7219

Our method with 2D U-Net

0.9318

0.8790

0.7701

0.6370

Our method with 3D U-Net

0.9370

0.8816

0.7835

0.6475

Our method with 2D nnU-Net

0.9530

0.9105

0.8255

0.7074

Our method with 3D nnU-Net

0.9574

0.9186

0.8400

0.7280

Back to article page