Abstract: An operation method of a brain amyloid PET processing system includes steps as follows. The whole brain white matter amyloid PET image is extracted from the smoothed amyloid PET image in the range of the whole brain white matter mask of the normalized brain space, and the uptake value with the preset maximum ratio in the whole brain white matter amyloid PET image is calculated; in the smoothed amyloid PET image of the normalized brain space, one or more voxels in the range of the whole brain gray matter mask are marked and counted, in which each voxel uptake value of the one or more voxels is greater than the uptake value of the preset maximum ratio of the whole brain white matter amyloid PET image, and the one or more voxels are used for interpretation training and test of the classification of the machine learning.

Abstract: A method is to be implemented by a computing device that stores a risk assessment model, and includes steps of: obtaining an entry of target physiological data and target magnetic resonance imaging (MRI) images of a brain of a subject with mild cognition impairment (MCI); obtaining, based on the target MRI images, voxel values respectively of primitive voxels that are related to grey matter of the brain of the subject; selecting, from among the primitive voxels, any primitive voxel satisfying a filtering criterion as a selected voxel; calculating an average of the voxel value(s) respectively of the selected voxel(s) to obtain an average target voxel value; and obtaining a probability of the subject developing Alzheimer's disease within a predetermined time period by feeding the average target voxel value and the entry of target physiological data into the risk assessment model.