A New Method for Diagnosing Alzheimer's Disease and Monitoring Its Severity Using FDG-PET

This study aimed to use fluorodeoxyglucose (FDG)-positron emission tomography (PET) images of the brain for the computer-aided diagnosis of Alzheimer's disease (AD) and monitor its severity using a new KNN and image similarity-measure-based approach. The first objective was to diagnose AD automatically using an image similarity index derived from mutual information. The second objective was to determine the association between the similarity measure and neuropsychological assessments. 95 patients with AD and 102 Cognitively Normal (CN) FDG-PET data were obtained from Alzheimer's Disease Neuroimaging Initiative (ADNI). We then found a distance metric indicative of the average similarity between any 3D image to CN and AD patients' images. The diagnosis is based on a threshold value for the distance. Then, the association of the distance metric values with Mini-Mental State Examination (MMSE) results were analyzed. The detection algorithm achieved an area under the ROC curve of 0.959 using a leave-one-out method. The correlation was 0.639 between MMSE and the distance metric. This is higher than previously reported values obtained with other indices. The new method does not need registration, segmentation, or feature extraction. Furthermore, no training is required for a growing or changing dataset.