Related Experiment Videos
Automated CT segmentation and analysis for acute middle cerebral artery stroke
J A Maldjian1, J Chalela, S E Kasner
1Departments of Radiology, Hospital of the University of Pennsylvania, Philadelphia, USA.
AJNR. American Journal of Neuroradiology
|June 21, 2001
Summary
This study developed an automated method to detect acute stroke hypodensity in the brain using CT scans. The algorithm accurately identified infarcts, improving upon initial clinical readings for middle cerebral artery stroke.
Area of Science:
- Neuroradiology
- Medical Imaging Analysis
- Computational Neuroscience
Background:
- Computed tomography (CT) scans possess quantitative properties suitable for automated analysis.
- Modern neuroimaging techniques can be applied to CT data for semiautomated analysis.
Purpose of the Study:
- To develop automated methods for analyzing CT scans.
- To identify hypodensity in the lentiform nucleus and insula in acute middle cerebral artery stroke patients.
Main Methods:
- Retrospective analysis of 35 CT scans (20 controls, 15 stroke patients).
- Data processing included interpolation, scalp stripping, atlas normalization, and anatomic segmentation.
- Voxel densities in target regions were compared to contralateral sides using statistical tests with spatial autocorrelation correction.
Main Results:
- Automated algorithm achieved excellent registration quality.
- The algorithm correctly identified most infarcts, with fewer false negatives than initial clinical readings.
- The automated method outperformed initial clinical interpretations in detecting stroke-related hypodensity.
Conclusions:
- An automated CT scan analysis method for detecting acute ischemia was developed.
- This approach shows potential for broader application in other brain regions and vascular territories.
- The method may assist in interpreting CT scans for hyperacute stroke cases.