Discrimination of normal aging, MCI and AD with multimodal imaging measures on the medial temporal lobe

This study aimed to compare the discrimination accuracy of hippocampal volume (HC-Vol), parahippocampal cingulum fractional anisotropy (PHC-FA), hippocampal glucose metabolism (HC-Glu), and any combination of the three measurements among normal control (NC), mild cognitive impairment (MCI), and Alzheimer's disease (AD). Three-dimensional MRI, diffusion tensor imaging, and FDG-PET were applied to age- and gender-matched 17 NC, 17 MCI, and 17 mild AD patients. Subjects also underwent a neuropsychological test battery including three verbal episodic memory tests. Logistic regression analyses were systematically conducted to select the best model for between-group discrimination. PHC-FA plus HC-Vol model, HC-Glu only model, and the model combining all three modalities were finally chosen for NC vs. MCI (discrimination accuracy: 79.4%), MCI vs. AD (73.5%), and NC vs. AD discrimination (94.1%), respectively. All the three imaging measures also showed significant correlation with all three episodic memory tests. These findings support that each imaging measure, respectively, and their combination have a stage-specific potential as a useful neuroimaging marker for detection and progression monitoring of early stage of AD.