Stroke: CT perfusion imaging can predict volume changes after cerebral infarction

The purpose of acute ischemic stroke treatment is to re-pattern the occluded arteries through intravenous thrombolysis or endovascular therapy (EVT), thereby saving the hypoperfused tissue

Over the years, the expected time from the onset of symptoms to treatment has determined the eligibility of these treatment options

Perfusion imaging, the most common of which is computed tomography perfusion (CTP), can help identify potentially salvable tissues (ie penumbra).

Based on the pre-determined mismatch pattern of core and perfusion lesion estimates, patients were selected for these trials

Machine learning is an alternative method to estimate the tissue fate of patients with ischemic stroke

The growth rate of infarction varies greatly from patient to patient and depends on the collateral circulation

In this way, Anke Wouters et al.

They trained a deep neural network to predict acute acuteness based on native computed tomography perfusion images, reperfusion time, and reperfusion status in a derived cohort (MR CLEAN trial [Multicenter Randomized Clinical Trial for Endovascular Treatment of Acute Ischemic Stroke in the Netherlands]) The final infarct volume of stroke patients

They also calculated the average absolute difference between the prediction of the deep learning model and the final infarct volume and the average absolute difference between the computed tomography perfusion imaging and the final infarct volume processed by RAPID software (iSchemaView, Menlo Park, CA)

They included 127 patients from MR CLEAN (derivation) and 101 patients from the CRISP study (validation)

The growth rate of individual infarcts is obtained, so that the final infarct volume can be estimated based on the time and grade of reperfusion

The core significance of this research is to verify a deep learning-based method , which improves the estimation of the final infarct volume compared with the classic computed tomography perfusion imaging process