A Quantitative Analysis in CTP images for Ischemic Stroke Lesion Segmentation

Due to the limitations of MR imaging availability, CT perfusion (CTP) has been used to determine the extension of stroke in early stages. But CT perfusion maps alone are not enough to segment the infarct core or penumbra precisely. So, the combination of noncontrast CT scans with CT perfusion maps would be accurate and efficient for the identification of the boundary of stroke lesions. In this paper, we aim to study the effect of reducing the perfusion maps number on the segmentation result and compare the performance of different CNN architectures such as Unet, ResUnet, and ResUnet++ using ISLES 2018 dataset.