VoxShield disrupts inter-slice frequency consistency and semantic logits in 3D medical images to degrade segmentation model performance to near-random levels with epsilon=4/255 perturbations.
In: Proceedings of the AAAI Conference on Artificial Intelligence (AAAI)
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 2roles
background 1polarities
background 1representative citing papers
NTGA is the first clean-label generalization attack under black-box settings but is vulnerable to adversarial training and image transformations, with newer attacks outperforming it.
citing papers explorer
-
VoxShield: Protecting 3D Medical Datasets from Unauthorized Training via Frequency-Aware Inter-Slice Disruption
VoxShield disrupts inter-slice frequency consistency and semantic logits in 3D medical images to degrade segmentation model performance to near-random levels with epsilon=4/255 perturbations.
-
SoK: A Comprehensive Analysis of the Current Status of Neural Tangent Generalization Attacks with Research Directions
NTGA is the first clean-label generalization attack under black-box settings but is vulnerable to adversarial training and image transformations, with newer attacks outperforming it.