Browsing by Author "Ingosan, Jeffrey S."

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    A survey of AI-based detection techniques for intracranial aneurysms
    (Institute of Electrical and Electronics Engineers, 2026) Dacanay, Kathrina T.; Ingosan, Jeffrey S.
    Early detection of intracranial aneurysms (IA) is essential for preventing subarachnoid hemorrhage, a condition associated with high morbidity and mortality. Recent advances in artificial intelligence (AI), particularly deep learning, have shown promising results in assisting radiologists across different angiographic modalities. This survey synthesizes findings from 23 peer-reviewed studies spanning CTA, TOF-MRA, DSA, 3D-RA, and multiphase CTA. Across modalities, CTA-based models achieved some of the highest reported patient-level sensitivities, often reaching ${95-100 \%}$, while multiphase CTA further improved sensitivity by $4-6 \%$ compared to single-phase scanning. TOF-MRA studies reported sensitivities ranging from $90-96 \%$, with segmentation models attaining Dice scores of $0.80-0.83$ and AUC values up to 0.96. DSA and 3D-RA systems demonstrated strong performance for larger aneurysms, achieving sensitivities of $91-97 \%$, though performance dropped for aneurysms $\leq 3$ mm. Across all modalities, 3D CNNs and U-Net variants were the most frequently used architectures, with two-stage detection-segmentation pipelines consistently reducing false positives. Despite these advancements, challenges remain, particularly false-positive rates in CTA, reduced sensitivity for very small aneurysms, and limited cross-institutional generalization. This review consolidates current trends, highlights persistent gaps, and outlines future directions for developing clinically robust AI systems for IA detection.