Artificial Intelligence Applications in Acute Ischemic Stroke

Acute ischaemic stroke results from arterial blockage that disrupts cerebral blood flow, leading to neuronal injury and long-term disability. Artificial intelligence has emerged across the stroke care continuum, from hyperacute diagnosis to prognostic modelling and workflow optimisation. In imaging, deep-learning architectures automate detection of large-vessel occlusions on computed tomography angiography, delivering results in under two minutes and expediting triage. Decision-support systems integrate clinical, demographic and imaging data to refine eligibility for reperfusion therapies such as intravenous thrombolysis and endovascular thrombectomy. Prognostic models leveraging machine-learning analyse perfusion imaging alongside baseline characteristics to predict functional outcomes, informing personalised rehabilitation strategies. The incorporation of secure messaging platforms linked to AI alerts has demonstrated measurable reductions in door-to-treatment times. Challenges in harmonising heterogeneous datasets, ensuring algorithm generalisability and maintaining model interpretability are being addressed through continuous data crowdsourcing and the development of explainability tools. These advances hold promise for scalable, global deployment of AI solutions that can reduce time to treatment, improve outcome prediction and ultimately lessen the worldwide burden of acute ischaemic stroke.

Research from Nature Portfolio

Recent studies have advanced automated detection of vessel occlusion using deep learning, demonstrating sensitivity in excess of 87 per cent and negative predictive values above 93 per cent across multi-centre external validation cohorts. These algorithms outperform existing regulatory-approved software by substantial margins in both sensitivity and predictive accuracy, and can process CT angiography datasets in under two minutes. An open-access imaging platform facilitates continuous algorithm refinement via data crowdsourcing, laying a blueprint for the development of generalisable AI tools in acute stroke diagnosis.

Artificial Intelligence Applications in Acute Ischemic Stroke publication trend

The graph below shows the total number of articles in artificial intelligence applications in acute ischemic stroke across all publications each year (not limited to Nature Index journals).

Technical terms

Acute ischaemic stroke: Sudden loss of brain function due to arterial occlusion and resultant tissue hypoxia.

Large-vessel occlusion (LVO): Blockage of a major cerebral artery, often targeted by endovascular therapies.

Computed tomography angiography (CTA): Imaging technique that visualises blood vessels using contrast-enhanced CT scans.

Convolutional neural network (CNN): Deep-learning model particularly suited to image recognition tasks.

Perfusion imaging: Assessment of blood flow dynamics in brain tissue, often via CT or MR perfusion scans.

Modified Rankin Scale (mRS): Ordinal scale measuring degree of disability or dependence following stroke.

References

1. Automated Large Vessel Occlusion Detection Software and Thrombectomy Treatment Times. JAMA Neurology (2023).
2. Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke. Nature Communications (2023).
3. Artificial Intelligence for Clinical Decision Support in Acute Ischemic Stroke: A Systematic Review. Stroke (2023).
4. Predicting stroke outcome: A case for multimodal deep learning methods with tabular and CT Perfusion data. Artificial Intelligence in Medicine (2023).