Classification of normal and abnormal knee joint using back-propagation neural network

Research output: Contribution to conferenceConference paperpeer-review

Abstract

The objective of this paper is to classify vibroarthrographic (VAG) signals, generated during joint movement, according to the pathological condition using time-frequency analysis and back-propagation neural network. VAG signals were segmented at 0.5 Hz by dynamic time warping (DTW) algorithm and Noise within the time-frequency distribution (TFD) of segmented VAG signals was reduced by singular value decomposition (SVD) algorithm. The features of VAG signals consist of the energy parameter (EP), the energy spread parameter (ESP), the frequency parameter (FP) and the frequency spread parameter (FSP) by Wigner-Ville distribution (WVD) and the magnitude pattern, the mean and the median frequency by fast Fourier transform (FFT). As a result, the average of the classification accuracy was 92.3 ±0.9 %. The proposed method showed good potential for non-invasive diagnosis of joint disorders.

Original languageEnglish
Title of host publicationProceedings of the 2008 International Conference on Bioinformatics and Computational Biology, BIOCOMP 2008
Pages483-488
Number of pages6
StatePublished - 2008
Event2008 International Conference on Bioinformatics and Computational Biology, BIOCOMP 2008 - Las Vegas, NV, United States
Duration: 2008.07.142008.07.17

Publication series

NameProceedings of the 2008 International Conference on Bioinformatics and Computational Biology, BIOCOMP 2008

Conference

Conference2008 International Conference on Bioinformatics and Computational Biology, BIOCOMP 2008
Country/TerritoryUnited States
CityLas Vegas, NV
Period08.07.1408.07.17

Keywords

  • Articular pathology
  • Back-propagation neural network
  • Dynamic time warping
  • Singular value decomposition
  • Vibroarthrography

Quacquarelli Symonds(QS) Subject Topics

  • Computer Science & Information Systems
  • Medicine
  • Data Science
  • Biological Sciences

Fingerprint

Dive into the research topics of 'Classification of normal and abnormal knee joint using back-propagation neural network'. Together they form a unique fingerprint.

Cite this