Support Vector Machine Prediction of Ischemic Tissue Fate in Acute Stroke Imaging

Shiliang Huang¹, Qiang Shen^1,2, Timothy Q. Duong^1,2

Predicting tissue outcome remains a challenge for stroke magnetic resonance imaging (MRI). In this study, a flexible support vector machine (SVM) algorithm was developed and applied to predict ischemic tissue fate on three stroke groups: 30-min, 60-min and permanent MCAO in rats. CBF, ADC and T2 were acquired during the acute phase up to 3hrs and again at 24hrs followed by histology. Infarct was predicted pixel-by-pixel using only acute (30-min) stroke data. Receiver-operating-characteristic analysis was used to quantify prediction accuracy. It was concluded that the SVM predictive model has the potential to serve as promising metrics for diagnosis, prognosis and therapeutic evaluation of acute stroke.