Region-specific agreement in ASPECTS estimation between neuroradiologists and e-ASPECTS software

Background and purpose: ASPECTS is a widely used measure of ischemic change on non-contrast CT. Although predictive of long-term outcome, it is limited by modest interobserver agreement. Machine learning strategies such as e-ASPECTS may help; here the authors compared e-ASPECTS with manual scoring by experienced neuroradiologists for all 10 individual ASPECTS regions.

Materials and methods: 178 baseline non-contrast CT scans from acute ischemic stroke patients undergoing endovascular thrombectomy were retrospectively reviewed by two independent neuroradiologists (with a third arbitrating) for a consensus read, scoring each ASPECTS region individually. All scans were then evaluated with e-ASPECTS (Brainomix), and interobserver agreement was calculated with kappa.

Results: Median ASPECTS was 9 for manual scoring and 8.5 for e-ASPECTS, with overall agreement κ=0.248. Regional agreement varied from κ=0.094 (M1) to κ=0.555 (lentiform). Prevalence-adjusted bias-adjusted kappa ranged from 0.483 (insula) to 0.888 (M3), with greater agreement for cortical areas.

Conclusion: Manual scoring and e-ASPECTS had fair agreement on a per-region basis, warranting further investigation using follow-up scans or MRI as the gold standard.