Cortical Thickness and Complexity in aMCI Patients: Altered Pattern Analysis and Early Diagnosis.

Abstract

Background: Amnestic Mild Cognitive Impairment (aMCI) is a prodromal phase of Alzheimer's disease. Although recent studies have focused on cortical thickness as a key indicator, cortical complexity has not been exhaustively investigated.

Objectives: To investigate the altered patterns of cortical features in aMCI patients and their correlation with memory function for early identification.

Methods: 25 aMCI patients and 54 normal controls underwent neuropsychological assessments and 3D-T1 MRI scans. Cortical thickness and complexity measures were calculated using CAT12 software. Differences between groups were analyzed using two-sample t-tests, and multiple linear regression was employed to identify features associated with memory function. A support vector machine (SVM) model was constructed using multidimensional structural indicators to evaluate diagnostic performance.

Results: aMCI patients exhibited extensive reductions in cortical thickness (pFDR-corrected <0.05), with complexity reduction predominantly in the left parahippocampal, entorhinal, rostral anterior cingulate, fusiform, and orbitofrontal (pFWE-corrected<0.05). Cortical indicators exhibited robust correlations with auditory verbal learning test (AVLT) scores. Specifically, the fractal dimension of the left medial orbitofrontal region was independently and positively associated with AVLT-short delayed score (r=0.348, p=0.002), while the gyrification index of the left rostral anterior cingulate region showed independent positive correlations with AVLT-long delayed and recognition scores (r=0.408, p=0.000; r=0.332, p=0.003). Finally, the SVM model integrating these cortical features achieved an AUC of 0.91, with 82.28% accuracy, 76% sensitivity, and 85.19% specificity.

Conclusion: Cortical morphological indicators provide important neuroimaging evidence for the early diagnosis of aMCI. Integrating multiple structural indicators significantly improves diagnostic accuracy.