CereBleed is an advanced deep learning tool designed for the automatic segmentation and volumetric quantification of intracranial hemorrhages, whether spontaneous or traumatic, on non-contrast CT scans.
The tool extracts detailed anatomical measurements and computes a severity index, integrating hemorrhage volume, and intracranial space distribution to provide a standardized, structured report.
Key Features
Automated Intracranial Hemorrhage Segmentation
Automatic segmentation of hemorrhage subtypes (intraparenchymal, intraventricular, subarachnoid, subdural, epidural).
Midline shift estimation
Midline shift estimation using deep learning-based landmark detection.
Severity Index Calculation
Mathematically derived severity index to reflect mass effect and clinical relevance.
Comprehensive Output
Structured and standardized reporting, optimized for integration into radiological workflows.
How it Works
CereBleed is built on state-of-the-art deep learning models for medical image segmentation, leveraging the power of nnUNet for robust, generalizable performance across different clinical datasets. While the research publication detailing this tool is currently under development, CereBleed is being validated through extensive experimentation as part of an ongoing research initiative.
Watch in Action
Publications
-
2025 medRxiv
Cepeda S. et al.
CEREBLEED: Automated quantification and severity scoring of intracranial hemorrhage on non-contrast CT
License
This software is distributed under the Creative Commons Attribution- NonCommercial (CC BY-NC) license. It is free to use, modify, and share for educational, personal, or non-profit purposes. Commercial use is strictly prohibited without prior written permission.