In a study involving 995 patients (mean age 69 years; 53% female), an AI-driven MRI model was evaluated for its ability to detect acute ischemic lesions. Case-based evaluation for detecting acute ischemic lesions showed 89% sensitivity (95% CI: 85–91%) and 90% specificity (95% CI: 87–92%) compared with neuroradiologist readings. No significant performance differences were observed across sex, age, or comorbidities, though specificity was reduced in cases with diffusion-weighted imaging artifacts. Multivariate analysis identified larger and fragmented ischemic lesions as predictors of higher sensitivity, while older lesion age was associated with reduced detection rates.
Apollo demonstrated demonstrated high accuracy in detecting acute ischemic lesions on MRI, performing nearly on par with experienced neuroradiologists. Sensitivity was influenced primarily by lesion characteristics, whereas specificity was affected by image quality.
Source: EUROPEAN JOURNAL OF RADIOLOGY