With only 1 to 7 labeled atlases, both multi-atlas and supervised deep learning segmentation approaches can reach near-upper-bound Dice scores on brain MRI, and a proposed semi-supervised registration variant improves boundary accuracy.
Multiresolution elastic matching
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Few Labeled Atlases are Necessary for Deep-Learning-Based Segmentation
With only 1 to 7 labeled atlases, both multi-atlas and supervised deep learning segmentation approaches can reach near-upper-bound Dice scores on brain MRI, and a proposed semi-supervised registration variant improves boundary accuracy.