AI-Powered Imaging Biomarkers Predict Recovery Outcomes in Acute Ischemic Stroke: Study Finds

Despite advances in stroke treatments, many patients still face long-term disabilities. Developing new therapies is challenging due to the complexity of measuring clinical outcomes. This study describes the potential of automated imaging biomarkers in predicting stroke recovery, offering a reliable and cost-effective tool for assessing treatment efficacy in clinical trials. Sonya Abraham, CSL Behring, King of Prussia, PA, United States, and colleagues aimed to validate an automated FIV algorithm by establishing the link between imaging biomarkers and clinical outcomes, supporting their use as imaging endpoints in clinical trials.

For this purpose, the researchers collected data from adult AIS patients who underwent mechanical thrombectomy and follow-up imaging within 12–96 hours of the initial assessment. Infarct volume was measured using non-contrast computed tomography and analyzed with AI-powered Brainomix 360 Stroke software. Key measures included total FIV, ischemic injury-corrected FIV (cFIV), hemorrhagic transformation (HT), anatomical distortion (AD) as an edema marker, and infarct growth (IG).

The primary clinical outcome was the modified Rankin Scale (mRS) score at 90 days, while the secondary outcome was the NIH Stroke Scale (NIHSS) score at 24 hours.

The key findings are as follows:

• A total of 843 out of 986 patients (85.5%) with complete data were included in the analysis. The median age of participants was 72 years, with 56.7% being male.
• The median baseline NIH Stroke Scale (NIHSS) score was 17. The median follow-up imaging time was 24 hours. The median NIHSS score at 24 hours was 11.
• At 90 days, 34% of patients achieved a modified Rankin Scale (mRS) score of 0–2.
• The median final infarct volume (FIV) was 30.2 mL.
• FIV showed a significant association with 90-day mRS (concordance = 0.819) and NIHSS score at 24 hours (concordance = 0.722).
• Other imaging biomarkers, including ischemic injury-corrected FIV (cFIV), hemorrhagic transformation (HT), anatomical distortion (AD), and infarct growth (IG), were also significantly linked to good clinical outcomes in both 90-day mRS and NIHSS at 24 hours.
• In multivariate analysis, FIV remained strongly associated with 90-day mRS outcomes.
• FIV displayed a bimodal distribution, indicating its relationship with the success or failure of recanalization during thrombectomy.

This study highlights the prognostic value of automated final infarct volume and its components—ischemic injury, hemorrhagic transformation, edema, and infarct growth—in predicting clinical outcomes after acute ischemic stroke.

"The findings support the use of AI-derived imaging biomarkers to enhance early-stage trials by refining efficacy assessment and understanding treatment mechanisms," the authors concluded.

Reference:

Abraham, S., Carone, D., Mielke, O., Heise, M., Swierczak, A., Bass, G., Gerry, S., Woodhead, Z. V., Namias, R., Garrard, J., Kallmes, D., Brinjikji, W., Vaclavik, D., Mikulenka, P., Nicholson, P., Thornton, J., Ford, G. A., & Harston, G. (2025). Automatically quantified follow-up imaging biomarkers predict clinical outcomes after acute ischemic stroke. Frontiers in Neurology, 16, 1483138. https://doi.org/10.3389/fneur.2025.1483138