Related Experiment Video
Updated: Jun 23, 2025
09:59
A Magnetic Resonance Imaging Protocol for Stroke Onset Time Estimation in Permanent Cerebral Ischemia
Published on: September 16, 2017
14.1K
Artificial intelligence for MRI stroke detection: a systematic review and meta-analysis.
Jonas Asgaard Bojsen1, Mohammad Talal Elhakim2, Ole Graumann3
1Research and Innovation Unit of Radiology, Odense University Hospital, University of Southern Denmark, Odense, Denmark. jabo@rsyd.dk.
Insights Into Imaging
|June 24, 2024
Summary
Artificial intelligence (AI) can accurately detect ischaemic stroke using magnetic resonance imaging (MRI) scans. Further research is needed to validate AI
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neurology
Background:
- Growing interest in AI for medical diagnostics.
- Limited understanding of AI's performance in MRI-based stroke assessment.
- Need for systematic evaluation of AI stroke detection tools.
Purpose of the Study:
- Assess AI's stroke detection performance in MRI.
- Identify reporting deficiencies in AI-related studies.
- Evaluate AI for both ischaemic and haemorrhagic stroke detection.
Main Methods:
- Systematic review and meta-analysis following PRISMA guidelines.
- Searched MEDLINE, Embase, Cochrane Central, and IEEE Xplore.
- Included studies using MRI in adults for stroke detection with AI.
- Assessed risk of bias using a modified QUADAS-2 tool and reporting quality with MI-CLAIM checklist.
Main Results:
- Included 33 studies; 15 had low risk of bias.
- Low-risk studies demonstrated better reporting quality.
- AI achieved 93% sensitivity and 93% specificity for ischaemic stroke detection.
- Limited evidence for AI's effectiveness in detecting haemorrhagic stroke.
Conclusions:
- AI technology shows high accuracy in detecting ischaemic stroke via MRI.
- Further validation is required for haemorrhagic stroke detection.
- Clinical utility of AI for stroke detection in MRI needs investigation.
