Related Experiment Video
Updated: Aug 22, 2025
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
42.7K
CMB-HUNT: Automatic detection of cerebral microbleeds using a deep neural network
Aleksandra Suwalska1, Yingzhe Wang2, Ziyu Yuan3
1Department of Data Science and Engineering, Silesian University of Technology, Akademicka 16, 44-100, Gliwice, Poland.
Computers in Biology and Medicine
|November 12, 2022
Summary
This study introduces CMB-HUNT, an AI tool for detecting cerebral microbleeds (CMBs) using only SWI MRI scans. CMB-HUNT achieves high accuracy and fewer false positives than existing methods, aiding in diagnosing small vessel diseases.
Area of Science:
- Neuroimaging
- Artificial Intelligence
- Medical Diagnostics
Background:
- Cerebral microbleeds (CMBs) are crucial biomarkers for cerebral small vessel diseases.
- Manual CMB detection is labor-intensive and error-prone.
- Current automated methods lack sufficient sensitivity, specificity, or require multiple MRI sequences.
Purpose of the Study:
- To develop an automated deep learning system (CMB-HUNT) for accurate CMB detection.
- To utilize only susceptibility-weighted imaging (SWI) data for CMB detection.
- To improve upon existing automated CMB detection techniques.
Main Methods:
- A deep neural network integrating SWI images and radiomic features was developed.
- The model, CMB-HUNT, was trained and validated on two independent patient datasets.
- Performance was evaluated using sensitivity and false positive rates.
Main Results:
- CMB-HUNT achieved 90.0% sensitivity on the hold-out test set with only 0.54 false positives per patient.
- External validation demonstrated 91.5% sensitivity and 1.9 false positives per patient, confirming generalization.
- The system's performance was comparable to existing studies, outperforming them in false positive reduction.
Conclusions:
- Deep learning models can effectively detect CMBs using solely SWI MRI data.
- CMB-HUNT offers a promising, accurate, and efficient solution for CMB detection.
- The system's ability to use a single MRI modality enhances its clinical applicability.
