Native-space UNet segmentation of subcortical structures outperforms template-space methods on held-out 7T data with Dice scores of 0.775 vs 0.713 for STN, but both struggle on 3T images.
Learning from spatially inhomogenous data: resolution-adaptive con- volutionsformultiplesclerosislesionsegmentation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Native space based pipelines outperform template space based pipeline in subcortical segmentation
Native-space UNet segmentation of subcortical structures outperforms template-space methods on held-out 7T data with Dice scores of 0.775 vs 0.713 for STN, but both struggle on 3T images.