Deep learning-driven MRI segmentation of choroid plexus volume: a novel biomarker for cognitive impairment in type 2 diabetes mellitus
摘要
Impaired glymphatic function has been linked to cognitive impairment in type 2 diabetes mellitus (T2DM). However, it remains unclear whether changes in choroid plexus volume (CPV), a critical component of the glymphatic system, are associated with cognitive impairment in T2DM. This study aimed to investigate the relationship between CPV changes and cognitive function, and identify factors independently associated with CPV alterations.
MethodsIn this prospective cohort study, 71 T2DM patients and 111 HC underwent 3T MRI scanning. CPV was automatically segmented from 3D T1-weighted images using the 3D nnU-Net deep learning model, trained and validated with manual annotations. Clinical data and cognitive performance were assessed. Cognitive tests included Montreal Cognitive Assessment (MoCA), Trail Making Test (TMT), Symbol Digit Modalities Test (SDMT), and Stroop Color and Word Test (SCWT). Statistical analysis included Mann-Whitney U tests, independent sample t-tests, partial correlation, and stepwise linear regression.
ResultsCPV was significantly increased in T2DM patients compared to HC (P = 0.018). Cognitive scores on the MoCA, TMT, SDMT, and SCWT significantly differed between groups (all P < 0.05). Stepwise regression analysis indicated that age, years of education, and CPV/eTIV ratio were independently associated with cognitive performance (TMT and SDMT scores; all P < 0.05). Furthermore, lower HDL cholesterol levels were associated with increased CPV, particularly in nonsmokers, nondrinkers, and individuals without hypertension (all P < 0.05).
ConclusionsEnlarged CPV may serve as a potential imaging biomarker of cognitive impairment in T2DM.