Classification of Knee Joint Vibroarthrographic Signals using Feedforward Neural Networks

S. Parsaeian (Iran), K. Tavakolian (Canada), M.B. Shamsolahi (Iran), and S. Rezaei (Canada)

Keywords

: Vibroarthographic signal (VAG),Feedforward Neural Network, Time-Frequency features.

Abstract

: Recording of Vibroarthrographic (VAG) signals, emitted during movement of knee joint is a noninvasive technique for diagnosis of cartilage disorders. In this research a total of 22 features from frequency, time and time-frequency and six autoregressive model coefficients have been extracted from VAG signal. These features were classified by a feedforward neural network to two normal and abnormal groups. For our first data set the classification accuracy was 87.5% and for the second dataset it was 75%.

Important Links:

Go Back