Table 3 Comparison of testing results with different methods
From: Weakly-supervised convolutional neural networks of renal tumor segmentation in abdominal CTA images
| Â | DSC | HD | ASD |
|---|---|---|---|
Constrained-CNN [21] | 0.705 | 102.178 | 8.271 |
Constrained-CNN [21] + 3D MC | 0.712 | 20.939 | 5.493 |
SDI [5] | 0.766 | 73.514 | 4.639 |
SDI [5] + 2D MC | 0.766 | 72.368 | 4.524 |
Ours (d = 5) | 0.820 | 37.633 | 3.879 |
Ours (d = 5) + 3D MC | 0.826 | 15.811 | 2.838 |
UNet [32] (Fully-supervised) | 0.849 | 84.69 | 4.886 |
UNet [32] (Fully-supervised) + 3D MC | 0.859 | 14.252 | 2.048 |