Textural content in 3T MR: an image-based marker for Alzheimer's disease

In this paper, we propose a study, which investigates the first-order and second-order distributions of T-2 images from a magnetic resonance (MR) scan for an age-matched data set of 24 Alzheimer's disease and 17 normal patients. The study is motivated by the desire to analyze the brain iron uptake in the hippocampus of Alzheimer's patients, which is captured by low T-2 values. Since, excess iron deposition occurs locally in certain regions of the brain. we are motivated to investigate the spatial distribution of which is captured by higher-order statistics. Based on the first-order and second-order distributions (involving gray level co-occurrence matrix) of T-2, we show that the second-order statistics provide features with sensitivity > 90% (at 80% specificity), which in turn capture the textural content in T-2 data. Hence, we argue that different text-tire characteristics of T-2 in the hippocampus for Alzheimer's and normal patients could be used as an early indicator of Alzheimer's disease.