Table 4.
Study | Our method | Mohammadi et al.18 | Jiang et al.16 | Zhang et al.17 | Yoganathan et al.19 |
---|---|---|---|---|---|
Radiation modality | CT-based brachytherapy, original vs edited auto-contour | CT-based brachytherapy, manual vs auto-contour | CT-based brachytherapy, manual vs auto-contour | CT-based brachytherapy, manual vs auto-contour | MRI-based brachytherapy, manual vs auto-contour |
DSC (%) | Bladder, 0.995 ± 0.003 | Bladder, 0.974 | Bladder, 0.860 ± 0.086 | Bladder, 0.869 ± 0.032 | Bladder, 0.90 ± 0.05 |
Rectum, 0.984 ± 0.097 | Rectum, 0.972 | Rectum, 0.858 ± 0.089 | Rectum, 0.821 ± 0.05 | Rectum, 0.76 ± 0.07 | |
Sigmoid, 0.964 ± 0.043 | Sigmoid, 94.1 | Sigmoid, 0.664 ± 0.123 | Sigmoid, 0.645 ± 0.079 | Sigmoid, 0.65 ± 0.12 | |
Small bowel, 0.987 ± 0.014 | Small bowel, 0.563 ± 0.129 | Small bowel, 0.803 ± 0.058 | Small bowel, 0.54 ± 0.12 | ||
Urethra, 0.868 ± 0.097 | |||||
MSD (mm) | Bladder, 0.052 ± 0.032 | - | - | - | - |
Rectum, 0.156 ± 0.238 | |||||
Sigmoid, 0.452 ± 0.870 | |||||
Small bowel, 0.269 ± 0.443 | |||||
Urethra, 0.474 ± 0.876 | |||||
95HD (mm) | Bladder, 0.399 ± 0.211 | Bladder, 1.64 | - | - | Bladder, 6.28 ± 3.42 |
Rectum, 1.005 ± 2.112 | Rectum, 1.37 | Rectum, 8.20 ± 4.07 | |||
Sigmoid, 3.215 ± 7.055 | Sigmoid, 1.84 | Sigmoid, 20.44 ± 11.70 | |||
Small bowel, 01.446 ± 2.480 | Small bowel, 22.3 ± 13.66 | ||||
Urethra, 2.447 ± 5.612 | |||||
HD (mm) | Bladder, 1.778 ± 0.610 | Bladder, 3.51 | Bladder, 19.981 ± 11.418 | Bladder, 12.1 ± 4.0 | - |
Rectum, 3.968 ± 5.322 | Rectum, 1.89 | Rectum, 12.273 ± 8.080 | Rectum, 9.2 ± 4.6 | ||
Sigmoid, 10.320 ± 14.330 | Sigmoid, 2.82 | Sigmoid, 98.409 ± 50.984 | Sigmoid, 19.6 ± 8.7 | ||
Small bowel, 10.177 ± 13.123 | Small bowel, 68.123 ± 33.781 | Small bowel, 27.8 ± 10.8 | |||
Urethra, 3.629 ± 7.499 |
Abbreviations: 95HD = 95% Hausdorff distance; CT = computed tomography; DSC = Dice similarity coefficient; MRI = magnetic resonance imaging; MSD = mean surface distance; HD= Hausdorff distance.
There have been limited studies evaluating the accuracy of auto-contouring in gynecologic brachytherapy. If a previous study compared multiple models, the most accurate model is reported. Measures reported in mean ± SD.