1
Return

RGCNN-nnUNet: Recurrent group equivariant nnU-Net for robust brain tissue segmentation on stroke NCCT

delete2026-06-27
delete0
PRE
AI
X
Xiang Li
F
Fengyuan Wang
S
Saurabh Bagchi
J
John Julius Volpi
T
Timea Hodics
S
Stephen T.C. Wong *
K
Kelvin Wong *
DOI：10.1016/j.compmedimag.2026.102792delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
• Equivariant kernels capture NCCT density changes, improving boundary segmentation. • Recurrent bottleneck cuts parameters by 60% compared to naive equivariant design. • Maintains high accuracy with just 25% training data, ideal for rare pathologies. • 5-fold ensemble segments full volumes in under 20s, aiding rapid decisions. • Accurate segmentation enables detection of subtle stroke cases often missed by experts.

Journal

Computerized Medical Imaging and Graphics cover
Computerized Medical Imaging and Graphics
IF:
4.9
Papers:
2.4K
Citations:
5.0K

Organization

P
Purdue University
Scholars:
2.6W
Papers: 2.0W
Citations: 147
H
Houston Methodist Research Institute
Scholars:
269
Papers: 104
Citations: 0
Cited Papers

Cited Papers

Citing Papers

Citing Papers