Table 4.
Segmentation evaluation by mIoU and mDice for each module combination when divided by the orientation of the heart axis.
| U-Net | YOLO | ED | VGG | mIoU | mDice | ||
|---|---|---|---|---|---|---|---|
| Apical | Non-Apical | Apical | Non-Apical | ||||
| ✓ | 0.1878 ±0.1097 | 0.1213 ± 0.0186 | 0.2697 ± 0.1410 | 0.1845 ± 0.0261 | |||
| ✓ | ✓ | 0.5793 ± 0.0315 | 0.4990 ± 0.0058 | 0.7064 ± 0.0405 | 0.6417 ± 0.0086 | ||
| ✓ | ✓ | ✓ | 0.5889 ± 0.0265 | 0.5210 ± 0.0160 | 0.7146 ± 0.0351 | 0.6653 ± 0.0140 | |
| ✓ | ✓ | ✓ | ✓ | 0.5855 ± 0.0167 | 0.5255 ± 0.0016 | 0.7114 ± 0.0264 | 0.6688 ± 0.0026 |
The values are the mean ± standard deviation of the three datasets for cross-validation. The apical group yielded higher values than the non-apical group. The addition of the YOLO significantly improved the segmentation accuracy, and the addition of the ED further improved it, irrespective of the cardiac axis orientation. The addition of the VGG contributed to the higher values in the apical group.