Table 4 Gross total resection versus residual tumor classification performances for both architectures, all input configurations, and over the validation and test sets.
From: Segmentation of glioblastomas in early post-operative multi-modal MRI with deep neural networks
Exp. | Data | Arch. | Patient-wise | ||
|---|---|---|---|---|---|
Sensitivity | Specificity | bAcc | |||
A | Val | nnU-Net | 99.81±0.35 | 2.53±2.21 | 51.17±1.22 |
AGU-Net | 79.70±6.69 | 68.01±10.43 | 73.86±4.94 | ||
Test | nnU-Net | 100.00 | 4.55 | 52.27 | |
AGU-Net | 84.31 | 63.64 | 73.98 | ||
B | Val | nnU-Net | 99.47±0.71 | 18.04±4.41 | 58.75±2.30 |
AGU-Net | 81.25±6.47 | 71.01±5.36 | 76.13±4.12 | ||
Test | nnU-Net | 98.04 | 45.45 | 71.75 | |
AGU-Net | 84.31 | 72.73 | 78.52 | ||
C | Val | nnU-Net | 99.81±0.35 | 5.64±3.44 | 52.73±1.76 |
AGU-Net | 79.29±10.08 | 74.00±11.13 | 76.64±4.87 | ||
Test | nnU-Net | 100.00 | 27.27 | 63.64 | |
AGU-Net | 78.43 | 90.91 | 84.67 | ||
D | Val | nnU-Net | 99.66±0.44 | 15.28±6.85 | 57.47±3.55 |
AGU-Net | 82.80±5.27 | 73.00±14.63 | 77.90±6.44 | ||
Test | nnU-Net | 100.00 | 40.91 | 70.45 | |
AGU-Net | 78.43 | 86.36 | 82.40 | ||
E | Val | nnU-Net | 100.00 | 6.12±4.30 | 53.06±2.15 |
AGU-Net | 85.61±4.83 | 72.63±9.39 | 79.12±4.60 | ||
Test | nnU-Net | 100.00 | 27.27 | 63.64 | |
AGU-Net | 86.27 | 77.27 | 81.77 | ||
