BloodNet improves ICH classification by modeling dependency between segmentation and classification tasks, reporting AUCs of 0.9493 and 0.9566 on held-out sets of over 1400 studies from more than 10 hospitals.
U-net: Convolutional networks for biomedical image segmentation,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Improved ICH classification using task-dependent learning
BloodNet improves ICH classification by modeling dependency between segmentation and classification tasks, reporting AUCs of 0.9493 and 0.9566 on held-out sets of over 1400 studies from more than 10 hospitals.