The automatic segmentation of the human amygdala in amnestic mild cognitive impairment
MetadataShow full item record
BACKGROUND: Mild cognitive impairment (MCI) is a clinical condition that is characterized by mild changes in cognition. The amnestic form of MCI (aMCI) primarily affects memory and is thought to represent a stage between healthy aging and Alzheimer’s disease (AD). The medial temporal lobe (MTL) and the limbic system are two areas of the brain that have been implicated in the amnestic form of MCI. While MCI represents a risk factor for AD, it does not always lead to dementias. Being a carrier of the APOE Ɛ4 allele has also shown to increase chances of progression from MCI to AD. OBJECTIVE: To determine whether the subnuclei of the amygdala, along with other specific regions within the MTL, can differentiate between cognitively normal individuals and age-matched subjects with aMCI. METHODS: T1-weighted magnetic resonance imaging (MRI) data from two sources, the Boston University Alzheimer’s Disease Center (BU-ADC) and the Alzheimer’s Disease Neuroimaging Initiative (ADNI), was compiled for cross-sectional analysis. 95 scans in total from 45 cognitively normal participants and 50 diagnosed with aMCI were analyzed and the volumes of interest were automatically generated by the developmental version of FreeSurfer v6.0. To evaluate how well the volumes could predict either group membership (i.e. control group or MCI group) or APOE Ɛ4 status (i.e. carrier or noncarrier), the variables were assessed by nominal logistic regression models. RESULTS: Six of the nine nuclei of the amygdala had significantly reduced volumes in the aMCI group compared to controls. The whole amygdala and the perirhinal cortex also demonstrated reduced volumes in the aMCI group compared to the control group. The whole amygdala was a good predictor of group membership (R2 = 0.1386, whole model test chi square = 18.21558, p = 0.0004), but none of the subnuclei were good predictors individually. A model containing the 9 nuclei, the entorhinal cortex, and the perirhinal cortex provided a good fit for predicting APOE Ɛ4 status fit (R2 = 0.3000, whole model test chi square = 36.29563, p = 0.0002) and the best predictor was the corticoamygdaloid transition area of the amygdala. CONCLUSIONS: The results of our study confirm previous findings of reduced whole amygdala volume and add to the limited literature of reduced perirhinal cortex and amygdaloid nuclei volumes in MCI compared to healthy controls. To the best of our knowledge, this was the first time the automatic segmentation atlas was used to analyze the volumes of nine subnuclei of the amygdala in a population of aMCI. Our model testing the volume of the whole amygdala accurately predicted aMCI subjects with 58% accuracy and controls with 70% accuracy; the accuracy rose to 69% when the entorhinal cortex and the perirhinal cortex were added to the model to predict aMCI subjects from controls. Additionally, the model for predicting APOE Ɛ4 status identified noncarriers of the allele at 85% accuracy. Future studies should consider increasing the sample size to better assess small ROIs and assess for differences in the separate hemispheres.