HTC-SGA Former is a hybrid CNN-Transformer model with MS-GLWA, SGFA, and BWACL that outperforms 14 prior methods on private coronary DSA datasets using only 0.81M parameters.
ChatGPT in healthcare: A taxonomy and systematic review
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
support 1representative citing papers
AMN fuses Swin Transformer and ResNet-50 via adaptive gating and trains with focal, boundary, and uncertainty losses to reach 0.82 mean Dice on the seven-class CoNIC benchmark.