In a side-by-side, real-case test, the deep learning model outperformed even expert stroke neurologists at detecting lacunar infarcts using only CTP, a scenario often faced in busy stroke units. This suggests AI could bolster confidence or efficiency in stroke triage, where diagnostics are notoriously error-prone and time-pressured. The real clinical payoff will be determined in live, prospectively monitored settings, not just retrospective datasets.
AI Outperforming Stroke Experts: Promise and Caution in Acute Lacunar Stroke Diagnosis
How does deep learning stack up against neurologist consensus for diagnosing acute lacunar stroke on CT perfusion—a notoriously difficult task, especially relevant for older adults?