Research Article
LCC-Net: A Lightweight Cross-Consistency Network for Semisupervised Cardiac MR Image Segmentation
Table 2
The impact of the number of the labeled subjects. The results are tested on the ACDC dataset.
| Labeled data | Method | LV (%) | RV (%) | MYO (%) | Mean (%) |
| 2 subjects | U-Net (baseline) | 76.1 | 24.7 | 69.1 | 56.6 | LCC-Net | 82.0 | 58.1 | 73.0 | 71.0 | 4 subjects | U-Net (baseline) | 79.6 | 51.7 | 71.1 | 67.5 | LCC-Net | 85.0 | 69.3 | 76.9 | 77.1 | 6 subjects | U-Net (baseline) | 83.4 | 62.3 | 74.0 | 73.2 | LCC-Net | 85.0 | 74.5 | 79.3 | 79.6 | 10 subjects | U-Net (baseline) | 82.1 | 70.1 | 76.6 | 76.3 | LCC-Net | 87.4 | 77.0 | 82.8 | 82.4 | Fully supervised (80 subjects) | U-Net (baseline) | 93.2 | 85.8 | 88.9 | 89.3 |
|
|