Feasibility and clinical value of visual MRI rating scales in differentiating dementia subtypes: correlation with montreal cognitive assessment

Abstract

Alzheimer’s disease (AD) and vascular dementia (VaD) are the two most common causes of dementia. Differentiation between them is clinically important, especially in settings with a high cerebrovascular disease burden. To assess the diagnostic utility of visual MRI rating scales—Medial Temporal Atrophy (MTA), Global Cortical Atrophy (GCA), and Small Vessel Disease (SVD)—in differentiating AD from VaD, and to examine their correlation with Montreal Cognitive Assessment (MoCA) scores. A cross-sectional case–control study was conducted at a tertiary care hospital between 2020 and 2022. Forty participants were included: VaD (n = 8), AD (n = 8), mixed dementia (n = 4), and controls (n = 20). MRI scans were evaluated using MTA, GCA, and SVD scales, with a second radiologist independently re-scoring all participants to assess inter-rater reliability. MoCA scores were obtained for clinical correlation, in preference to the MMSE, given its greater sensitivity to vascular and executive cognitive deficits. Statistical analysis included ANOVA with Bonferroni post hoc tests and eta-squared effect sizes, Spearman correlation, ROC curve analysis, weighted Cohen’s kappa, and multivariable logistic regression (Firth’s penalized method). MTA scores were significantly higher in AD and mixed dementia compared to controls (p < 0.001, η2 = 0.76). SVD scores were significantly higher in VaD and mixed dementia compared to controls (p < 0.001, η2 = 0.81). No significant differences were observed for GCA. MoCA correlated negatively with both MTA and SVD (rho = − 0.4, p < 0.05). ROC analysis demonstrated strong predictive performance of SVD in identifying VaD (AUC 0.85). Inter-rater reliability was almost perfect for all three scales (weighted κ 0.86–0.99). A multivariable logistic regression model combining MTA, GCA, SVD, and age was significant overall (χ2(4) = 22.40, p < 0.001) in differentiating AD from VaD, though individual predictor estimates were unstable given the small sample. Visual MRI rating scales are simple, reproducible, and reliable tools for dementia assessment that can be scored on routine MRI without specialized sequences or post-processing. MTA is more useful for AD, whereas SVD strongly supports VaD, and their joint diagnostic value is reinforced by a significant multivariable model. Their correlation with the more sensitive MoCA reinforces their role in clinical evaluation, and reliable, accessible imaging-based differentiation of AD from VaD can support earlier, more appropriate treatment selection in routine practice.