Recognize Vital Features for Classification of Neurodegenerative Diseases

Abstract

Neurodegenerative diseases including Parkinson diseases, Huntington diseases, ataxia, myoclonus and amyotrophic lateral sclerosis are medically, hereditarily, pathologically fluctuated and are described by its symptoms and execution of motor impairment. Early diagnosis, efficient treatment planning and observation of PD and other NDDs are achieved by gait dynamic characterization. In this study, the data samples are of 15 subjects with PD, 20 subjects with HD disease, 13 subjects with ALS and 16 subjects of healthy or fit persons. The database is composed of one minute recording of force sensitive resistor (FSR) Signal. Both left and right stride-stride footfall contacts are obtained from FSR signals. For feature deduction, two-level wavelet decomposition is done using discrete wavelet transform (DWT). The acquired features were evaluated using the means of 10-trials for fivefold cross-validation (FFCV) in LDA with a random forest classifier (RFC). In the result, the NDD pathologies are detected by the proposed method. For NDD location, the random forest classifier gives the better outcome contrast with SVM and QB ordinary classifiers. The experiment proves that the accuracy, sensitivity and the specificity of the proposed system is highly accurate and efficient than the previous methods. The percentage of the proposed method is 98.24, 97.92 and 96.78% as each.