Prediction of Cerebral Hyperperfusion Syndrome After Carotid Revascularization Using Deep Learning

Cerebral hyperperfusion syndrome (CHS) was initially described as a clinical complication following carotid endarterectomy (CEA), but it may occur after both CEA and carotid artery stenting. It is characterised by throbbing ipsilateral frontotemporal or periorbital headache, and sometimes diffuse headache, eye and facial pain, vomiting, confusion, macular oedema, visual disturbances, focal motor seizures with frequent secondary generalisation, focal neurological deficits, and intracerebral or subarachnoid haemorrhage.

Knowledge of CHS among physicians remains limited. Most studies report an incidence of 1-3% after carotid endarterectomy. CHS is most common in patients with increases of more than 100% in cerebral perfusion compared with baseline after carotid revascularization, and is rare in patients with perfusion increases of less than 100% compared with baseline.

The pathophysiological mechanism of CHS is only partially understood. The chronic low-flow state induced by severe carotid disease results in compensatory dilation of cerebral vessels distal to the stenosis, as part of the normal autoregulatory response to maintain adequate cerebral blood flow (CBF). In this chronically dilated state, the vessels lose their ability to autoregulate vascular resistance in response to changes in blood pressure. Dysautoregulation has been shown to be proportional to the duration and severity of chronic hypoperfusion. After revascularization and reperfusion, impaired cerebral autoregulation may contribute to a cascade of intracranial microcirculatory changes, with an inability to respond adequately to the augmentation of CBF following carotid recanalization.

Although most patients present with mild symptoms and signs, progression to severe and life-threatening complications can occur if CHS is not recognised and treated promptly. Because CHS is diagnosed on the basis of several non-specific signs and symptoms, patients may be misdiagnosed as having one of the better-known causes of perioperative complications, such as thromboembolism.