Abstract Background and aims To establish artificial intelligence-derived, Magnetic Resonance Imaging (MRI)-based etiological subtypes of cerebral small vascular disease (CSVD). Methods A multicenter, cohort study that enrolled 1,036 adults with mild cognitive impairment before September 31, 2025 with the first month follow-up was completed. MRI radiomic features were extracted using an artificial intelligence driven segmentation and clustering framework for etiological subtyping. The primary outcome was the clinical risks of the imaging-driven MCI subtypes. Results The analysis revealed three CSVD subtypes based on seven imaging features: a lacune-specific subtype, characterized by severe lacunes, was associated with lower high-density lipoprotein cholesterol (HDL-C) (mean SD, 0.9 0.2 mmol/L; OR, 3.27; 95% CI, 1.41-5.60), higher systolic blood pressure (SBP), and lower apolipoprotein A1 (mean SD, 1.2 0.3 g/L; OR, 1.01; 95% CI, 1.01-1.71); a white matter hyperintensities (WMHs)-specific subtype, marked by more severe WMHs, was associated with elevated SBP, normal HDL-C (mean SD, 1.1 0.2 mmol/L), and apolipoprotein A1 (mean SD, 1.3 0.2 g/L) levels; and a hippocampal-atrophy subtype, marked by significant hippocampal atrophy, was associated with fewer vascular risk factors (HDL-C, mean SD, 1.2 0.4 mmol/L; apolipoprotein A1, mean SD, 1.3 0.3 g/L). Visuospatial and language deficits were prominent in the lacune- and WMHs-specific subtypes. These clusters were validated in 1130 patients from the ADNI database. Abnormal HDL-C metabolism was linked to specific gene mutations. Conclusions This study identified and validated three distinct imaging subtypes of CSVD, providing a framework for precise diagnosis and targeted therapeutic strategies. Conflict of interest Jinjie Liu. nothing to disclose Figure 1 - belongs to Methods Table 1 - belongs to Results
Building similarity graph...
Analyzing shared references across papers
Loading...
Jinjie Liu
Sibo Liu
D J Li
European Stroke Journal
Dalian University of Technology
Dalian Municipal Central Hospital
Building similarity graph...
Analyzing shared references across papers
Loading...
Liu et al. (Fri,) studied this question.
www.synapsesocial.com/papers/69fd7e90bfa21ec5bbf06dfd — DOI: https://doi.org/10.1093/esj/aakag023.507