Feasibility of an AI-assisted transcranial duplex sonography protocol for early detection of intracerebral haemorrhage: the HYPER-AI-SCAN single-centre prospective study

INTRODUCTION: Intracerebral haemorrhage (ICH) is associated with high early mortality and morbidity. Early clinical deterioration is common and influenced by haematoma expansion, which can occur within the first hours after symptom onset. Transcranial duplex sonography (TCD) is a rapid, non-invasive tool that may aid in early ICH detection but is highly operator-dependent. Artificial intelligence (AI)-based analysis of ultrasound images has shown promise in other fields but has not yet been validated in acute ICH. METHODS AND ANALYSIS: This is a single-centre, prospective feasibility study involving 500 patients with acute ischaemic and haemorrhagic stroke (<48 hours from onset), with a 1:4 haemorrhagic-to-ischaemic ratio reflecting population prevalence. Patients with infratentorial haemorrhage will be excluded. Once computed tomography (CT) confirms the diagnosis, TCD will be performed, and coded sonographic data will be collected. AI models, including pre-trained convolutional neural networks and transformer-based architectures, will be fine-tuned using sonographic images labelled by CT diagnosis. The model will aim to classify binary outputs: 'ICH suspected' versus 'No ICH'. Clinical, radiological and temporal variables will be recorded to evaluate associations with outcomes. ETHICS AND DISSEMINATION: Ethical approval has been obtained. Informed consent will be collected. Data will be coded and stored securely. Results will be disseminated through peer-reviewed journals and conferences. TRIAL REGISTRATION NUMBER: Not applicable at this stage (observational AI study).