Identification of EMCI in MR Brainstem Structure Using Fractal Measures and Random Forest Approach.

In this work, an attempt has been made to assess the texture variations of brainstem occurring in the Early Mild Cognitive Impairment (EMCI) condition. Fractal dimension is calculated for the segmented brainstem volume using Higuchi's and Detrending moving average (DMA) method. These measures are validated using Random forest classifier. DMA shows better performance when compared to Higuchi's method in categorising EMCI stage. Thus, the proposed approach using brainstem image's DMA fractal measures with random forest can be used in diagnosis of EMCI.