Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
August 26, 2025Frontiers in NeuroscienceOpen Access

Automated ischemic stroke lesion detection on non-contrast brain CT: a large-scale clinical feasibility test AI stroke lesion detection on NCCT

View Full Paper
Ask AI
Bookmark
Share

Authors

JHJoonNyung HeoYonsei UniversityWRWi‐Sun RyuE Ink (South Korea)JCJong‐Won ChungChonnam National University

Discussion

Loading...

Member takes

Implication

Multicenter cohort study demonstrates that automated ischemic stroke lesion detection on NCCT provides reliable outcomes, implying potential clinical utility.

Key Points

  • Automated detection of ischemic lesions on non-contrast CT achieved 75.3% sensitivity and 79.1% specificity, enhancing diagnostic reliability.
  • In a cohort of 603 patients undergoing endovascular thrombectomy, NCCT-derived lesion volumes correlated with DWI follow-up volumes, showing ρ = 0.60.
  • The modified 3D U-Net model used data from 2,214 patients to assess clinical feasibility and external validation in 458 subjects.
  • Higher lesion volumes were linked to increased hemorrhagic transformation rates and poorer outcomes, highlighting the model's prognostic relevance.

Cite This Study

Heo et al. (2025) studied this question.

synapsesocial.com/papers/68af7e047567bf4f94ff5784https://doi.org/10.3389/fnins.2025.1643479
View Full Paper
Ask AI
Bookmark
Share