Machine Learning 2025-03-27
Computer Vision in Emergency Radiology
Advanced 3D convolutional neural networks for improved detection of intracranial bleeding in CT imaging, focusing on emergency radiology applications where rapid and accurate hemorrhage detection is critical for patient outcomes.
5C Network Research Team · arXiv · DOI: 10.48550/arXiv.2503.20306
Key Findings
- 3D CNNs outperform 2D slice-based approaches for detecting intracranial hemorrhage subtypes
- Rapid inference time enables real-time triage in emergency department workflows
- System identifies critical bleeds within seconds, supporting faster clinical decision-making
- Multi-class detection covers subdural, epidural, subarachnoid, intraparenchymal, and intraventricular hemorrhage
