Purpose: To investigate the subtle functional connectivity alterations of aMCI based on AAL atlas with 1024 regions (AAL_1024 atlas). Materials and methods: Functional MRI images of 32 aMCI patients (Male/Female: 15/17, Ages: 66.8 +/- 8.36 y) and 35 normal controls (Male/Female:13/22, Ages: 62.4 +/- 8.14 y) were obtained in this study. Firstly, functional connectivity networks were constructed by Pearson's Correlation based on the subtle AAL_1024 atlas. Then, local and global network parameters were calculated from the thresholding functional connectivity matrices. Finally, multiple-comparison analysis was performed on these parameters to find the functional network alterations of aMCI. And furtherly, a couple of classifiers were adopted to identify the aMCI by using the network parameters. Results: More subtle local brain functional alterations were detected by using AAL_1024 atlas. And the predominate nodes including hippocampus, inferior temporal gyrus, inferior parietal gyrus were identified which was not detected by AAL_90 atlas. The identification of aMCI from normal controls were significantly improved with the highest accuracy (98.51%), sensitivity (100%) and specificity (97.14%) compared to those (88.06%, 84.38% and 91.43% for the highest accuracy, sensitivity and specificity respectively) obtained by using AAL_90 atlas. Conclusion: More subtle functional connectivity alterations of aMCI could be found based on AAL_1024 atlas than those based on AAL_90 atlas. Besides, the identification of aMCI could also be improved. (C) 2018 Published by Elsevier B.V.