This post-hoc analysis evaluates hypodensity volume's role in EVT outcomes for patients with large ischemic cores, suggesting implications for treatment decisions.
Key Points
This study aims to determine how deep learning segmentation of ischemic core hypodensity relates to treatment outcomes in patients undergoing endovascular treatment.
Post-hoc analysis of SELECT2 trial data
Measured hypodense volumes on NCCT using DL and manual segmentation
Assessed agreement between DL and manual segmentations with concordance correlation coefficients
Used Zou's Modified Poisson regression to analyze associations with ambulation outcomes
Total hypodense volume agreement between DL and manual segmentation was 0.73 (CCC), while severely hypodense volume was 0.90 (CCC)
Both total and severely hypodense volumes were linked to lower independent ambulation rates after EVT
Severely hypodense volumes were identified as significant indicators of reduced EVT benefit