PET image classification using HHT-based features through fractal sampling

Abstract

Medical image classification is currently a challenging task that can be used to aid the diagnosis of different brain diseases. Thus, exploratory and discriminative analysis techniques aiming to obtain rep- resentative features from the images, play a decisive role in the design of effective Computer Aided Diagnosis (CAD) systems, which is spe- cially important in the early diagnosis of dementias. In this work we present a technique that allows extracting discriminative features from Positron Emission Tomography (PET) by means of an Empirical Mode Decomposition-based (EEMD) method. This requires to transform the 3D PET image into a time series which is addressed by sampling the image using a fractal-based method which allows to preserve the spa- tial relationship among voxels. The devised technique has been used to classify images from the Alzheimer's Disease Neuroimaging Initiat- ive (ADNI) achieving up to a 90.5% accuracy in a differential diagnosis task (AD vs. controls), which proves that the information retrieved by our methodology is significantly linked to the disease.