Back to all papers

Gray matter brain age predicts cognitive outcome one year after ischemic stroke.

May 26, 2026pubmed logopapers

Authors

Tseng WI,Hsieh YC,Hsu YC,Huang LK,Fu CK,Chen DY,Chen JH,Hong CT,Lu YH,Chan L,Chiou HY

Affiliations (8)

  • AcroViz, Inc., Taipei, Taiwan, (R.O.C).
  • Ph.D. Program in Medical Neuroscience, College of Medical Science and Technology, Taipei Medical University, Taipei, (R.O.C.).
  • Department of Neurology, Shuang Ho Hospital, Taipei Medical University, New Taipei City, (R.O.C.).
  • Department of Neurology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, (R.O.C.).
  • Department of Radiology, Shuang Ho Hospital, Taipei Medical University, New Taipei City, (R.O.C.).
  • Department of Radiology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, (R.O.C.).
  • School of Public Health, College of Public Health, Taipei Medical University, Taipei, (R.O.C.).
  • Institute of Population Health Sciences, National Health Research Institutes, Miaoli County, (R.O.C.).

Abstract

BackgroundPredicting post-stroke cognitive impairment (PSCI) remains challenging.ObjectiveThis study validated two brain age metrics-Gray Matter Brain Age (GMBA) and Predicted Age Difference (PAD)-as independent predictors of 12-month cognitive decline following ischemic stroke.MethodsWe analyzed 39 patients with ischemic stroke. Brain age was estimated using a machine learning model trained on healthy controls (n = 362). Baseline assessments, including demographics, clinical severity, and thick-sliced (7 mm) 2D T1- and T2-weighted MRI, were performed during acute hospitalization (median 5 days post-stroke). Change in Clinical Dementia Rating-Sum of Boxes (<b>Δ</b>CDR-SB) was measured between baseline and 12 months. Multiple linear regression (MLR) and receiver operating characteristic (ROC) analysis identified predictors of <b>Δ</b>CDR-SB.ResultsMLR identified two significant models. In Model 1, PAD (β = 0.109, p = 0.035) and chronological age (β = 0.082, p = 0.038) were independent predictors (Adjusted R<sup>2</sup> = 0.183). In Model 2, GMBA emerged as the sole robust predictor (β = 0.092, p = 0.002; Adjusted R<sup>2</sup> = 0.201). ROC analysis showed GMBA possessed the highest discriminative ability (AUC = 0.717, p = 0.026), followed by PAD (AUC = 0.691, p = 0.050), while chronological age was not significant (AUC = 0.629, p = 0.188).ConclusionsBoth GMBA and PAD are independent predictors of PSCI. Notably, these metrics maintained high predictive utility even when derived from real-world, thick-sliced MRI protocols. Integrating brain age estimation into routine clinical workflows can potentially facilitate early identification of patients at high risk for PSCI.

Topics

Journal Article

Ready to Sharpen Your Edge?

Subscribe to join 11k+ peers who rely on RadAI Slice. Get the essential weekly briefing that empowers you to navigate the future of radiology.

We respect your privacy. Unsubscribe at any time.