FILTER FEATURE SELECTION ANALYSIS TO DETERMINE THE CHARACTERISTICS OF DEMENTIA

Abstract

Dementias are known as neuropsychiatric disorders. Two-dimensional sliced brain scans can be generated via magnetic resonance imaging. Three-dimensional measurements of regions can be reached from those scans. Numerical brain features can beextracted through operating the Freesurfer tool. Parametrizing those featuresand demographic informationin learning algorithms can label an unknown sample as healthy or dementia. On the other hand, some of the features in the initial set may be less practical than others. In this research, the aim is to decrease theinputfeature-count, a total of 2939 attributes,as a first step to determine the mostdistinctive dementia characteristics. To that end, a total of 2264 ADNI dataset samples (471 AD, 428 lMCI, 669 eMCI, 696 healthy controls) are divided into two sets: 65% training set (1464 samples) and 35% test set (800 samples). Variousfilter feature selection algorithms(Information Gain,Gain Ratio,Symmetrical Uncertainty, Pearson’s Correlation, Correlation-based Feature Subset Selection)are tested over different parameters together withmultipleBayesian-based and tree-basedclassifiers. Test performance accuracy rates upto 76.50% are analyzed in detail.Instead of processingthewholefeatureset, the overall performance tends to increase with correctly fewer attributes taken.

Keywords

• alzheimer’s disease
• dementia
• filter feature selection
• adni
• freesurfer