A Landmark-Guided Dual-Stream Synergistic Framework for Automated Intracranial Aneurysm Detection in Magnetic Resonance Angiography
A deep learning system detected brain aneurysms on routine MRI with 87% sensitivity without needing complex preprocessing, though performed worse on tiny aneurysms.
A landmark-guided dual-stream deep learning framework for intracranial aneurysm detection on TOF-MRA (n=1055 scans) achieved lesion-wise sensitivity of 0.87 with 1.23 false positives per case, without requiring impractical vessel segmentation. Performance was anatomically robust but declined for small (≤3mm) aneurysms, which are clinically challenging.
What the study was
- Study design
- Retrospective development and validation study
- Population
- Patients with TOF-MRA scans (n=1055)
- Sample size
- 1055
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- Journal of Imaging Informatics in Medicine
Why it surfaced
Solid n=1055 dataset for AI-based IA detection; landmark-guided approach reduces annotation burden; clinically meaningful performance on a high-stakes vascular condition.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.