The Role of Artificial Intelligence in Stroke Imaging in Emergency Settings: A Systematic Review
Bibliographic record
Abstract
Rapid and accurate interpretation of neuroimaging is critical in acute stroke, but variability among human readers and the urgency of clinical workflows pose major challenges. Artificial intelligence (AI) has emerged as a promising adjunct in emergency stroke imaging, with the potential to enhance detection, scoring, and prognostication. We systematically reviewed the role of AI in this context, focusing on diagnostic performance, workflow feasibility, and implementation across key imaging modalities. A systematic search of PubMed, Scopus, Web of Science, and Cochrane CENTRAL was conducted from inception to August 20, 2025, following Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. Eligible studies were original English-language research that applied AI to emergency stroke imaging. Data were extracted on study design, population, imaging modality, AI architecture, performance metrics, workflow aspects, and interpretability, with study quality assessed using the Checklist for Artificial Intelligence in Medical Imaging (CLAIM). Nine studies met the inclusion criteria. AI models achieved high accuracy for intracranial hemorrhage (ICH) detection on non-contrast computed tomography (NCCT) scans, with area under the curve (AUC) values up to 0.98. Real-world analyses reported balanced accuracy around 0.93 with near-real-time processing. Automated Alberta Stroke Program Early CT Score (ASPECTS) grading demonstrated almost perfect agreement with expert consensus (κ up to 0.90), outperforming individual radiologists in the hyperacute phase. Ischemic lesion detection using convolutional neural networks (CNNs) applied to magnetic resonance imaging (MRI) and computed tomography angiography (CTA) achieved accuracies of 83-86%. Symmetry-based methods further improved performance, though limitations were noted in posterior circulation strokes. Prognostic models integrating imaging and clinical data yielded moderate-to-good performance (AUC 0.79-0.91), with multimodal deep learning outperforming single-modality or clinical-only models. Workflow studies reported AI processing times of 2-4 minutes, although data transfer and system integration remained key bottlenecks. Interpretability tools, such as Gradient-weighted Class Activation Mapping (Grad-CAM) and Shapley Additive Explanations (SHAP), have enhanced transparency in several studies. Overall, AI demonstrates strong diagnostic and workflow potential in emergency stroke imaging, particularly for ICH detection, automated ASPECTS, and large vessel occlusion (LVO) alerts. Multimodal and transformer-based approaches show promise for outcome prediction and lesion segmentation, but further external validation and seamless integration into clinical workflows are required. AI is best positioned as a supportive tool to augment, rather than replace, clinical expertise in acute stroke care.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.013 | 0.060 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.007 | 0.008 |
| Bibliometrics | 0.010 | 0.009 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".