Table 6.
The segmentation results and 95% confidence intervals for the 15 multiple runs of original U-Net, U-Net with HF, and U-Net with average-pooled labels (AVG). Bold fonts indicate the best result.
DSC | HD95 (mm) | AVD (%) | recall | F1-score | ||
---|---|---|---|---|---|---|
| ||||||
Original U-Net | Mean±SD | 0.8033±0.0204* | 6.2509±1.6870* | 18.8337±6.086* | 0.7261±0.0538* | 0.7273±0.0466* |
95% CI | [0.0060, 0.0089] | [−0.7435, −0.3917] | [−2.8959, −1.2721] | [0.0212, 0.0440] | [0.0228, 0.0298] | |
Original + AVG | Mean±SD | 0.8037±0.0191* | 6.9537±2.5907* | 17.6740± 6.4449 | 0.7373±0.0552* | 0.7340±0.0513* |
95% CI | [0.0055, 0.0085] | [−1.7264, −0.8143] | [−2.0247, −0.1762] | [0.0086, 0.0341] | [0.0126, 0.0266] | |
Original + HF | Mean±SD | 0.8107±0.0203 | 5.6833±1.8080 | 16.7497±6.9472 | 0.7587±0.0456 | 0.7536±0.0436 |
p-value corrected by FDR < 0.05 between U-Net with HF and other (paired t-test).
Abbreviation – SD: standard deviation; CI: confidence interval.