SPECT image classification based on NMSE feature correlation weighting and SVM

Abstract

This paper shows a new computer aided diagnosis (CAD) technique for the early Alzheimer's disease (AD) based on single photon emission computed tomography (SPECT) image feature selection and a statistical learning theory (SLT) classifier. Conventional evaluation of SPECT is time consuming, subjective and prone to error because images often rely on manual reorientation, visual reading of tomographic slices and semiquantitative analysis of certain regions of interest (ROIs). The study proposed is carried out in order to find the ROIs and the most discriminant image parameters with the aim of reducing the dimensionality of the input space, thus improving the accuracy of the system. This innovative method consists of voxel-based Normalized Mean Square Error (NMSE) feature extraction, a t-test with feature correlation weighting for feature selection and support vector machine (SVM) for image classification. Among all the features evaluated, coronal standard deviation and sagittal correlation parameters are found to be the most effective ones for reducing the dimensionality of the input space and improving the diagnosis accuracy when a linear kernel SVM is used. The proposed method yields an up to 98% classification accuracy, thus outperforming recent developed methods for early AD diagnosis including the 78.5% accuracy of the classical baseline voxel-as-features (VAF) approach.