From: Automated meniscus segmentation and tear detection of knee MRI with a 3D mask-RCNN
Author(s) and years | Data size | Method(s) | Performance |
---|---|---|---|
Aldrin et al. [24] & 2017 | 18 | 2D U-net | Diceā=ā0.752 Sensitivityā=ā0.796 |
Norman et al. [25] & 2018 | 638 | 2D U-net | Dice for medial meniscusā=ā80.9% Dice for lateral meniscusā=ā75.3% |
Tack A et al. [26] & 2018 | 552 | 3D U-net | Dice for medial meniscusā=ā83.8% Dice for lateral meniscusā=ā88.9% |
Our 3D Mask-RCNN | 546 | 3D Mask-RCNN | Dice for bilateral meniscusā=ā92.4% |