fidoptRibSeg v2, individual ribs
A native chest-abdomen CT with its 24 ribs numbered one by one (RibSeg v2): the fixed images of RibSeg's registration pairs.
Read before comparing (3)
- Split. The 30 test and 29 development cases inherit RibSeg's patient split. Only the native fixed image and its manual or manually corrected masks are used; the moving image is not counted as a new segmentation case.
- Files. The images and annotations are referenced from RibSeg, not copied.
- Ribs. The task segments whole ribs and their numbering; RibFrac's fracture annotations do not stand in for rib masks. The source cohort includes fractures.
| Accuracy | Cost | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| # | Method | Dice | Surface Dice | HD95 mm | Time s | RAM GB | GPU GB | Machine | Images |
| 1 | KonfAI segmentation app · TotalSegmentator (total), the authors' packageGPUapp | 0.805 ±0.036 | 0.875 ±0.038 | 9.87 ±3.02 | 25.8 | 5.1 | Workstation | ||
| 2 | KonfAI segmentation app · TotalSegmentator (total), KonfAI appGPUapp | 0.790 ±0.039 | 0.859 ±0.040 | 11.01 ±3.04 | 3.0 | 6.8 | Laptop | ||
| 3 | CT-FM whole-body segmentationGPUapp | 0.573 ±0.066 | 0.602 ±0.061 | 39.24 ±7.32 | 164 | 16.1 | 6.8 | Workstation | |
| 4 | MONAI wholeBody CTGPUapp29 failed | 0.014 ±0.028 | 0.015 ±0.031 | 24.47 | 4.8 | 8.3 | Workstation | ||
The order averages each method's ranks on Dice, Surface Dice, HD95 mm, so the first one is not always the best in every column. ± is the 95% interval of the mean over the 30 cases. Two neighbours are tied when a paired Wilcoxon test on the shared cases does not find the upper one better on Dice (5%): the places a method could hold are then shown under its own. The best value of each column is in colour. Click a row to see its images, a column title to sort by it.
1 result not ranked: an older run, cases still missing, or a failure on every case
| Accuracy | Cost | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| # | Method | Dice | Surface Dice | HD95 mm | Time s | RAM GB | GPU GB | Machine | Images |
| MONAI wholeBody CTGPUappolder run | 0.691 ±0.053 | 0.743 ±0.055 | 17.18 ±2.97 | 22 | 3.2 | 14.3 | Laptop | ||
Dice: how much the organ contours overlap, 1 is perfect. Surface Dice: the share of the contours' surfaces within a tolerance of each other, 1 is perfect. HD95: distance between the contours' surfaces.
Case by case
Click a score to look at that result: the image, the truth in green and the method's map in red. Underlined scores have their image; for the others you see the case without the result.
About the data
- Truth
- contours by a program, verified by people The RibSeg v2 morphological or point-assisted rib annotations, with manual correction and numbering by people.
- Source
- github.com/HINTLab/RibSeg
doi.org/10.1109/TMI.2023.3313627
arxiv.org/abs/2210.09309
docs.google.com/spreadsheets/d/1lz9liWPy8yHybKCdO3BCA9K76QH8a54XduiZS_9fK70
zenodo.org/records/3893508
zenodo.org/records/3893498
zenodo.org/records/3893496
zenodo.org/records/3993380
ribfrac.grand-challenge.org
arxiv.org/abs/2402.09372
doi.org/10.1016/j.ebiom.2020.103106 - Licence
- CC BY-NC 4.0 (the CTs); the labels carry no licence of their own
29 cases for tuning, never tested
14_638/fixed, 137_449/fixed, 570_199/fixed, 26_585/fixed, 628_66/fixed, 649_302/fixed, 114_582/fixed, 128_609/fixed, 298_363/fixed, 555_154/fixed, 537_291/fixed, 616_90/fixed, 106_476/fixed, 430_270/fixed, 20_601/fixed, 253_278/fixed, 51_533/fixed, 72_348/fixed, 98_189/fixed, 460_575/fixed, 442_489/fixed, 650_426/fixed, 134_211/fixed, 469_31/fixed, 610_80/fixed, 73_172/fixed, 63_68/fixed, 75_381/fixed, 514_339/fixed