Comparing deep learning stroke segmentation in NCCT, CTA, and CTP: Accuracy, domain transfer, and temporal sampling effect

Key Points

• This research aims to evaluate the performance of automated stroke lesion segmentation across different CT acquisition methods.
• Evaluated nnU-Net for stroke lesion segmentation
• Compared accuracy across NCCT, CTA, and CTP
• Investigated the effects of multitimepoint imaging on segmentation performance
• Multitimepoint imaging shows significantly better segmentation accuracy than NCCT and CTA
• Automated segmentation quality varies across different CT acquisition types