Combining Non-Contrast CT Radiomics with MR PREDICTS Improves Prediction of Futile Recanalization after Mechanical...

Area of Science:

• Neurology
• Radiology
• Artificial Intelligence

Background:

• Mechanical thrombectomy (MT) is a key treatment for acute ischemic stroke (AIS) from large vessel occlusion (LVO).
• Futile recanalization (FR), defined by poor functional outcome despite successful reperfusion (eTICI 3), occurs in a significant patient subset.
• Predicting FR and post-MT outcomes remains challenging.

Purpose of the Study:

• To assess the efficacy of baseline non-contrast CT (NCCT) radiomics in predicting functional and imaging outcomes post-MT.
• To determine if combining NCCT radiomics with MR PREDICTS variables enhances prediction accuracy for FR and early ischemic changes.

Main Methods:

• Retrospective analysis of 167 AIS-LVO patients achieving eTICI 3 reperfusion after MT.
• Extraction of 377 radiomic features from baseline NCCT slices.
• Development and evaluation of ensemble machine learning (EML) models using radiomics alone and combined with MR PREDICTS variables.

Main Results:

• NCCT radiomics alone showed limited predictive ability for FR (AUC 0.54).
• Integration with MR PREDICTS improved FR prediction (AUC 0.69, accuracy 0.71), though not statistically significant.
• Both radiomics-only and combined models demonstrated good prediction for 24-hour NCCT ischemic changes (AUCs 0.73 and 0.78, respectively).

Conclusions:

• Baseline NCCT radiomics effectively predict early imaging evolution post-MT.
• Combining NCCT radiomics with MR PREDICTS offers improved, though not significantly different, prediction of futile recanalization.
• Further multicenter validation is necessary to confirm these findings for clinical application.